Status of insecticide resistance in Anopheles gambiae (s.l.) of The Gambia

Background Vector control activities, namely long-lasting insecticidal nets (LLIN) and indoor residual spraying (IRS), have contributed significantly to the decreasing malaria burden observed in The Gambia since 2008. Nevertheless, insecticide resistance may threaten such success; it is important to regularly assess the susceptibility of local malaria vectors to available insecticides. Methods In the transmission seasons of 2016 and 2017, Anopheles gambiae (s.l.) larvae were sampled in or around the nine vector surveillance sentinel sites of the Gambia National Malaria Control Programme (GNMCP) and in a few additional sampling points. Using WHO susceptibility bioassays, female adult mosquitoes were exposed to insecticide-impregnated papers. Molecular identification of sibling species and insecticide resistance molecular markers was done on a subset of 2000 female mosquitoes. Results A total of 4666 wild-caught female adult mosquitoes were exposed to either permethrin (n = 665), deltamethrin (n = 744), DDT (n = 1021), bendiocarb (n = 990) or pirimiphos-methyl (n = 630) insecticide-impregnated papers and control papers (n = 616). Among the 2000 anophelines, 1511 (80.7%) were Anopheles arabiensis, 204 (10.9%) Anopheles coluzzii, 75 (4%) Anopheles gambiae (s.s.), and 83 (4.4%) An. gambiae (s.s.) and An. coluzzii hybrids. There was a significant variation in the composition and species distribution by regions and year, P = 0.009. Deltamethrin, permethrin and DDT resistance was found in An. arabiensis, especially in the coastal region, and was mediated by Vgsc-1014F/S mutations (odds ratio = 34, P = 0.014). There was suspected resistance to pirimiphos-methyl (actellic 300CS) in the North Bank Region although only one survivor had the Ace-1-119S mutation. Conclusions As no confirmed resistance to bendiocarb and actellic 300CS was detected, the national malaria control programme can continue using these insecticides for IRS. Nevertheless, the detection of Ace-1 119S mutation warrants extensive monitoring. The source of insecticide pressure driving insecticide resistance to pyrethroids and DDT detected at the coastal region should be further investigated in order to properly manage the spread of resistance in The Gambia. Electronic supplementary material The online version of this article (10.1186/s13071-019-3538-0) contains supplementary material, which is available to authorized users.

Background Malaria control in The Gambia as in other endemic countries relies on vector control with long-lasting insecticidal nets (LLIN) and indoor residual spraying (IRS) [1]; globally, these two interventions have averted about 78% of malaria-related deaths and morbidity [2]. Nevertheless, as they rely on efficacious insecticides, the emergence and spread of insecticide resistance may represent a major threat [3][4][5], although LLINs would still offer personal protection against mosquito bites [6].
IRS has been implemented in The Gambia since 2008, and the insecticide employed for this purpose has varied over the years, firstly DDT and then bendiocarb in 2015-2016, after significant resistance to DDT was observed [4,7], and pirimiphos-methyl (actellic 300CS) since 2017. Changing the insecticide used for IRS follows the recommendations of the World Health Organization (WHO) Global Plan for Insecticide Resistance Management (GPIRM) of regularly rotating insecticides of different classes [8].
The choice of the insecticide for use in IRS should be based on the susceptibility of the local vectors to the WHO approved insecticides [9]. The Gambia has routinely conducted insecticide susceptibility tests to guide IRS campaigns (GNMCP unpublished reports). The most recent country-wide survey was done in 2013-2014 and reported high levels of resistance to DDT and pyrethroids [4].
In collaboration with the Gambian National Malaria Control Programme (GNMCP), the susceptibility of the local malaria vectors to the insecticides commonly used for vector control, namely deltamethrin, permethrin, DDT, bendiocarb and actellic 300CS, was determined during the 2016 and 2017 malaria transmission seasons.

Study area
This is a nationwide study carried out in villages at 9 GNMCP vector surveillance sentinel sites: Essau, Farafenni, Mansa Konko, Kuntaur, Brikama, Basse, Bwiam, Pirang, Bakau. Additional sites, Besse, Yallal Ba, Chogen Wellingara, Wali Kunda, Gambisarra, Gunjur Koto and Sare Wuro, that formed part of a countrywide insecticide susceptibility study in 2013-2014, were also included to provide a robust map of insecticide susceptibility patterns [4] (Additional file 1: Table S1). Villages were grouped into 7 regions to provide an overview of susceptibility per region (Fig. 1). Kanifing Municipality (KMC) located in West Coast Region (WCR) and representing an urban settlement, was considered a region on its own. Central River Region (CRR) was also divided into North (CRRN) and South (CRRS) bank. Similarly, Upper River Region (URR) was divided into North (URRN) and South (URRS). North Bank Region (NBR) was taken as one region.

Collection and rearing of larvae
Malaria vectors were collected over two malaria transmission seasons (July-December), in 2016 and 2017. Two mosquito collection strategies were used to generate adults for susceptibility tests following the WHO susceptibility bioassay protocol [10], namely larva sampling in their natural breeding habitats or (when larvae could not be found) collection of blood-fed or gravid adults.
Anopheles gambiae (s.l.) larvae were collected from breeding sites in all but two villages, Ngedden and Sare Seedy, where blood-fed or gravid mosquitoes were sampled using mouth aspirators. The larvae were reared according to standard conditions of 27 ± 2 °C and relative humidity of 80 ± 5% in the insectaries at the MRC  [4] field stations in Basse and Wali Kunda. The blood-fed/ gravid females were placed in a cage and induced to lay eggs by placing a dump filter paper in a Petri dish. They were fed on 10% sugar solution until no more eggs were laid. The eggs were reared to larvae under similar conditions as larvae collected from the field until adults emerged.

Insecticide susceptibility bioassays of Anopheles gambiae (s.l.)
Insecticide-impregnated papers were prepared and supplied by Liverpool Insecticide Testing Establishment (LITE) at the Liverpool School of Tropical Medicine, Liverpool, UK. To validate the effectiveness of the WHO test papers, susceptible laboratory mosquito colony (Anopheles coluzzii Yaoundé strain) was exposed to all the impregnated and control papers. Groups of 20-25 three-to five-days-old adult female mosquitoes were exposed for an hour to either 4% DDT, 0.25% pirimiphos-methyl, 0.1% bendiocarb, 0.75% permethrin and 0.05% deltamethrin impregnated papers [10]. For each insecticide class exposure experiment, random samples of field-collected and susceptible laboratory strain were separately exposed to appropriate control papers. Mortality 24 h post-exposure was recorded for each replicate experiment. All exposure experiments were performed under the temperature of 26 ± 2 °C and relative humidity of 70 ± 10%.
Since mosquito density was low, it was not possible to perform the recommended 5 replicates per insecticide per village. Therefore, to increase the power of the mortality estimates for each insecticide, village data were pooled according to geographical location into 7 regions ( Fig. 1, Additional file 1: Table S2). The temperature and humidity were also recorded during the exposure periods.
All exposed mosquitoes were stored in labelled individual Eppendorf tubes with a desiccant-silica gel. They were then transported to the coast for molecular genotyping at the MRC to determine species [11,12]; known insecticide resistance mutations at the Voltage-gated sodium channel (Vgsc-1014F and Vgsc-1014S) gene that confer knockdown resistance (kdr) against DDT and pyrethroids and mutation at the Ace-1 (Ace-1-119S) gene conferring resistance to organophosphates and carbamates were also investigated using Taqman polymerase chain reactions (PCR) [13,14]. This was done in a subset of 2000 mosquitoes randomly selected according to 4 main criteria: insecticide exposed; phenotypic status (dead or alive); year of collection and lastly, the number of samples selected per region was proportional to the number of mosquitoes collected (Additional file 1: Table S3).

Statistical analysis
All statistical analyses were performed in R statistical package [15]. Mortality was estimated by dividing the number of dead mosquitoes per replicate 24 h postexposure by the total number of mosquitoes exposed. Spearman's correlation was used to test the effect of temperature and humidity on mortality. Odds ratios, Chi-square and Fisherʼs exact tests were used to measure the association between resistance mutations and susceptibility to insecticides. Proportionality tests (tests of confidence intervals) [16] were used to establish differences in mortality between regions for each of the insecticides tested. General linear models (GLM) with logit link function for a binomial dependent variable was used to model the effect of kdr, insecticide, species, region and year on mortality.

Results
A total of 4666 female An. gambiae (s.l.) were successfully reared and tested for phenotypic resistance. Of these, 665 were tested against permethrin, 744 to deltamethrin-, 1021 to DDT-, 990 to bendiocarb-, 630 to actellic 300CS-impregnated papers and 616 to control papers. In addition, 600 laboratory strains from Yaoundé, Cameroon, were also used as controls during exposure bioassays.

Anopheles gambiae (s.l.) phenotypic resistance to insecticides
All susceptible laboratory strain mosquitoes died upon exposure to insecticide-impregnated papers except those in the control papers. During the exposure experiments, mortality in the control group was always less than 5% (data not shown) and therefore Abbottʼs correction was not applied to the mortality estimates. There was no effect of temperature and humidity on mortality, P = 0.299.

Deltamethrin and permethrin
There was marked resistance to deltamethrin and permethrin, especially in KMC and WCR in the coastal region, with only about 30% mortality (Fig. 3). Resistance was also observed in NBR and CRR in 2016 but not in 2017 (Fig. 3). In URRS, resistance to pyrethroids was low in both years with mortality rate of 97%. Permethrin was not tested in 2017 in WCR and NBR. There was high Spearmanʼs correlation coefficient (r = 0.89, P = 0.03) between deltamethrin and permethrin susceptibility. Paired and unpaired Mann-Whitney U-tests did not detect differences in mortalities between the two insecticides.

DDT
Higher phenotypic resistance to DDT was recorded in KMC and WCR in the coastal regions, with mortality of about 30% (Fig. 4). In NBR, resistance was low in 2016, with 96% mortality but mortality was significantly lower  in 2017, 57.5% (P < 0.001). In CRRS, CRRN and URRS, resistance was only suspected in 2016; however, in 2017 resistance in URRS increased significantly, P = 0.0001 (Fig. 3).

Frequency of insecticide resistance molecular markers
Of the 2000 An. gambiae (s.l.) screened for sibling species and markers of insecticide resistance, 93.65% (1873) were successfully genotyped for resistance mutations. Three mosquitoes (2 from CRRN and one from NBR) had the Ace-1 G119S mutation but only one survived exposure to carbamates and no further statistical testing was done.
The frequency of the Vgsc-L1014F (kdrF) and Vgsc-L1014S (kdrS) alleles varied by species and region, with the latter significantly higher than the former, P < 0.001 (Additional file 1: Figure S1 and Table S4). The kdrS and kdrF frequencies were significantly higher at the coast (WCR and KMC) than in the other regions (P < 0.001).  In An. arabiensis, the frequency of the kdr alleles varied significantly by region, P < 0.01. Low numbers for other species precluded statistical estimation of frequency and confidence intervals around the estimates (Additional file 1: Table S4).

Knockdown resistance (kdr) markers and phenotypic resistance
In WCR and KMC, An. arabiensis with the kdr mutations were more likely to survive after exposure to either pyrethroids or DDT in both 2016 (odds ratio, OR = 34, P = 0.014) and 2017, (OR = 2.5, P < 0.025). In URR, kdrS conferred survival advantage for An. arabiensis exposed to pyrethroids/DDT only when the data from the two years were pooled (OR = 2.5, P = 0.003). In CRR, there was no relationship between survival and any of the kdr mutations.
In a generalized linear model (GLM), kdrF/S, together with region and year explained significant variation in mortality estimates of An. gambiae (s.l.) to pyrethroids and DDT (Table 1). Although species was not significantly associated with mortality, the best-fit model was the one including species as a variable. Including insecticide did not significantly improve the model, and this variable was excluded. Both kdrF and kdrS significantly explained survival of mosquitoes against the pyrethroids and DDT (Table 2).

Discussion
In The Gambia, malaria vectors are still susceptible to bendiocarb and actellic 300CS although one An. arabiensis from CRRN and two An. gam-An. col hybrids from CRRN and NBR had the Ace-1 119S mutation that confers resistance to organophosphates and carbamates. Conversely, resistance to pyrethroids and DDT, mediated by Vgsc-L1014F and Vgsc-L1014S, was widespread and more frequent in the coastal region in western Gambia.
Consistent with other published reports [4,17,18], three members of the An. gambiae complex, namely An. gambiae (s.s.), An. arabiensis and An. coluzzii, including An. gam-An. col hybrids, were identified and their distribution varied substantially by region; An. arabiensis was the dominant species in almost all regions while An. gambiae (s.s.) was found mainly in two regions. Such results differ from those of the latest nationwide entomological study carried out in 2013-2014 as An. gambiae (s.s.) and An. coluzzii were much more frequent at that time [4].
Such change in species composition may be due to the use of bendiocarb and actellic 300CS for IRS, which killed the more anthropogenic and endophilic An. gambiae (s.s.) and An. coluzzii, increasing the proportion of the more zoophilic An. arabiensis. This change in species composition has also been observed in Kenya, Tanzania and Senegal [19][20][21]. The predominance of An. arabiensis compared to other, more efficient vectors such as An. gambiae (s.s.), could have also contributed to the decline in malaria transmission observed in The Gambia over the last few years.
Insecticide resistance is known to vary among species and in time and changes in species composition may impact on the status of insecticide resistance [4]. Differences in species composition in NBR could have led to observed variation in observed insecticide resistance.

Mechanisms of insecticide resistance
Vgsc-1014F and Vgsc-1014S mutations in An. arabiensis were associated with phenotypic resistance, particularly in WCR and KMC. This differs from previous reports from The Gambia in which the lack of association between mutations and resistance in this species was explained by the few An. arabiensis tested and the lack of kdr survivors [4,22]. However, in neighbouring Senegal, kdr alleles in