Effect of Bendiocarb Indoor Residual Spraying on Entomological Inoculation Rate of Anopheles arabiensis in Northwestern Highlands of Ethiopia

Entomological inoculation rate (EIR) is a method to estimate the level of human exposure to infective mosquito bites and assess impacts of vector control measures. The objective is to assess the effect of indoor residual spray (IRS) on blood meal index (BMI), sporozoite infection rate (SR), and EIR in An. arabiensis under local ecological settings in Ethiopia. A total 1541 fresh fed (FF) female An. arabiensis collected by CDC light trap and PSC were processed at the Center for Disease Control and Prevention Laboratory, Atlanta, Georgia, USA, to determine their BMI and SR, using enzyme-linked immunosorbent assay (ELISA). IRS reduced the abundance of FF female An. arabiensis in sprayed villages (n=62) while the number remained high in non-sprayed villages (n=1,690). The relative adjusted reduction in human blood feeding index (HBI) due to IRS varied between 3 and 10% except in 2014 when no human blood was detected in any of the three mosquitoes tested. The relative adjusted reduction in P. falciparum infection and EIR in An. arabiensis was 100% after IRS. The results illustrated that IRS was strong enough to reduce EIR in An. arabiensis . IRS is recommended to control malaria transmission in areas of similar ecological set.


Introduction
Current malaria vector control strategies rely heavily on indoor residual spraying (IRS) and long-lasting insecticide-treated mosquito nets (LLINs). The impact of these intervention tools on entomological malaria transmission risk factors needs to be evaluated. The level of exposure to infective mosquito bites could be measured using entomological inoculation rate (EIR) in the vector [8,28]. The EIR is defined as the number of infective bites received by an individual per unit time (night, month, or year). It is the product of human-biting rate (HBR) and plasmodium sporozoite infection rate (SR) [26,47].
The human landing catch (HLC) is the most commonly used method to determine the human-biting rate because it is the direct measure of human-vector contact [47]. However, due to ethical and logistic constraints associated with HLC, light trap catches (LTC), pyrethrum spray catches (PSC), and exit trap catches could be used as alternatives to human landing catches [26] to estimate the HBR. In this study, the Centers for Disease Control and Prevention (CDC) light trap and PSC mosquito sampling methods were used to estimate the HBR.
Malaria is a public health problem in Ethiopia. Indoor residual spraying and LLNs are the frontline pillars of malaria vector intervention tools that have been used in all malarious parts of the country. However, studies on the impact of these interventions on EIR are either limited or unavailable [2,41]. Besides, EIR varies from region to region, even from locality to locality. Therefore, narrowing this knowledge gap would be valuable for vector control program. The present study was carried out to assess the impact of the current vector control strategy specifically IRS on BMI, SR, and EIR.
Dichlorodiphenyltrichloroethane (DDT) was the choice of insecticide for IRS operation that had been used for decades in many malarious areas of Ethiopia except at a few places where malathion was used for DDT-resistant vector populations. This was continued until 2007 when DDT was replaced by deltamethrin due to the development of DDT resistance in the major malaria vector populations [14]. Payable to the occurrences of deltamethrin resistance in different vector populations, in view of the possibility of cross-resistance between DDT and pyrethroid insecticides and the scaling up of the distributions of pyrethroidtreated LLINs, IRS control program again replaced deltamethrin by bendiocarb (carbamate group) in 2010 and still in use for IRS operations in different parts of the country.
The residual efficacy of bendiocarb with the recommended concentration could last between 2 and 6 months depending on the nature of sprayable surfaces [12]. Therefore, bendiocarb was the choice of insecticide used for IRS operation during the present study.

Study site
The study was carried out in two adjacent villages, namely, Andassa (N11° 30′ 14.6″, 037 o 29′ 27.8″) and Tikurit (11° 30′ 49.8″, 037 o 28′ 02.8″), Bahir Dar Zuria District, North West Ethiopia. These villages were separated by Andassa River and buffered by about 2 km vegetable and fruit farms. The study villages were selected purposively by considering malaria endemicity and the history of IRS implementation. Indoor residual spraying and LLINs are the primary intervention tools that have been used for years against A. arabiensis (important vector of the study area). The vector has developed different levels of insecticide resistance to insecticides of different classes recommended for both LLIN treatment and IRS operation [1].

Design
A comparative study was carried out in Andassa and Tikurit villages. The study was conducted for 2 consecutive years. Andassa received two rounds of sprays one in 2013 and another in 2014, while no spray was implemented in Tikurit. Participants in the unsprayed villagers were provided with treated bed nets free of charge, and individuals found infected received free treatment at the nearest health center. The susceptibility status of A. arabiensis to bendiocarb was confirmed before the application of IRS.

Mosquito sampling
Adult female A. arabiensis were collected from 24 residential houses (12 houses/village and 6 houses/sampling method) using pyrethroid spray catch (PSC) and CDC light trap sampling methods. PSC was applied by spraying pyrethroid insecticide under which a white muslin cloth was placed to facilitate knocked-down mosquito collection. Human baits, sleeping in beds covered with treated bed nets, were used to reinforce CDC light traps. Mosquitoes that were collected by each sampling method before and after IRS were then stored individually in tubes containing silica gel to process and determine their BMI and SR in the lab.

Blood meal host source and sporozoite rate determination
Enzyme-linked immunosorbent assay (ELISA) originally described by Beier et al. [5] and CS-ELISA [46] protocols were adopted and used for BMI and SR analyses, respectively. Blood-fed mosquitoes preserved individually in tubes containing silica gel were used to determine their BMI and SR. Heads-thoraxes of mosquitoes were separated from their abdomens, and each body part (abdomen/ head-thorax) was given a corresponding ID number and kept individually in tubes for analyses.

Blood meal source determination
The mosquito abdomen, which was kept individually in tubes containing silica gel, was ground in a tube containing 100 μl of phosphate buffer saline (PBS) with a plastic pestle fitted with foot-operated grinder. The pestle was rinsed twice with 200 μl of PBS to achieve the final volume of 500 μl. The samples were either incubated at room temperature for 3 h and then stored at 4°C and tested the next day. Mosquitoes were tested to assess the blood meal origin of human and bovine only because these hosts were the predominant hosts of the vector during the study period. A 96-well ELISA plate was used, and 50 μl of the positive control for the blood meal host being tested was loaded. Wells A2-A5 had 50 μl of the negative controls, and wells A6-A8 were blanks containing 50 μl of blocking buffer. The plate was then covered and incubated for 3 h. The mosquito triturate was then aspirated by multichannel pipet, and the plate was washed three times with 200 μl PBS-Tween20 (5%). For a full 96-well plate, the peroxidase conjugate anti-host IgG antibody was prepared by adding 4800 μl of blocking buffer and 19.2 μl of anti-host and 1 μl of 1:100,000 of each of the negative control [5]. Fifty microliter of peroxidase conjugate was added to each well, and the plate was covered and incubated for 1 h at room temperature. The plate was then washed three times with 200 μl PBS-Tween20 (5%), and the one component ABTS peroxidase substrate was added to each well. PBS-Tween20 was aspirated by multichannel pipet, and plates were banged between washes. After 30 min of covered incubation at room temperature, the plate was read with the SpectraMax 340 plate reader (Molecular Devices) at 414 nm.

Sporozoite rate determination
The head-thorax of a mosquito, which was kept individually in step tubes containing silica gel, was ground in 1.5 μl microcentrifuge grinding tube containing 50 μl PBS with a plastic pestle fitted with foot-operated grinder. The pestle was rinsed twice with 100 μl of PBS and dried with tissue paper to prevent contamination between mosquito samples.
A 96-well ELISA PVC plate was coated with 50 μl of capture monoclonal antibodies (mAb) of each plasmodium sporozoite species (Pf, Pv-2010, and Pv-247) in each well of the ELISA plates (a separate plate used for each species), covered and incubated for half an hour. After the well contents were aspirated, plates were banged upside down on paper towel five times. The wells were then filled with 200 μl blocking buffer (BB), covered with lid and incubated for 1 h at room temperature. Well contents are aspirated and the plate is banged on paper towel five times. Samples and controls were loaded into the plate (well 1A, positive control; wells 1B-1H, negative control; and the rest of the wells with mosquito triturate) and covered and incubated for 2 h. Well contents were aspirated, and the plates were banged upside down on paper towel five times and washed two times with 200 μl of PBS-Tween20. The wells were aspirated, and plates were banged upside down five times with each wash. Then a 50 μl of peroxidase conjugate solution of each plasmodium sporozoite species (Pf, Pv-2010, and Pv-247) was added to each well and covered and incubated for 1 h. After aspirating the well contents and banging the plates, wells were washed three times with 200 μl of PBS-Tween20 and aspirated, and plates were banged five times with each wash. Finally, a 100 μl of the substrate solution was added per well, covered with cover plate and incubated for 30 min. The results were then read visually at the SpectraMax 340 plate reader (Molecular Devices) at 405-414 nm. All positive samples were retested for confirmation.

Determination of entomological inoculation rate
Plasmodium EIR of A. arabiensis was determined based on CDC light traps and PSC. The EIR was estimated from PSC samples as described by the World Health Organization [48] using the formula: number of fresh fed (FF) mosquitoes caught by PSC/ no. human occupants who spent the previous night in sprayed house) × (number of human fed mosquitoes/number of mosquitoes tested for human blood meal) × (number of sporozoite positive ELISAs/ number of mosquitoes tested, i.e., HBR × CSP rate. The human-biting rate was calculated by dividing the total number of freshly fed A. arabiensis caught in PSC by the total number of occupants who slept in the houses in the previous night of mosquito collection and multiplied by the HBI. The HBI was calculated as the proportion of Anopheles mosquitoes that fed on humans to the total Anopheles analyzed for blood meal origin [18,19]. EIR from CDC light trap catches was estimated using the standard method, 1.605 (number of circumsporozoite-positive ELISA results from CDC light trap/ no. of mosquitoes tested) × (number of mosquitoes collected by CDC LT/ no. of CDC LT catches), and the alternative method, 1.605 (no. positive ELISA/no. catches) [13].

Data analyses
The relative adjusted reduction in human blood feeding index (HBI), sporozoite rate (SR), and the entomologic inoculation rate (EIR) of the vector after intervention was calculated using the formula [ C2 and T1 and T2 describe the either the number of A. arabiensis or percentages of BMI, SP, or EIR in sprayed (T) and non-sprayed villages (C) before IRS (subscript 1) and after IRS (subscript 2). This formula takes into account that changes in the mosquito population and parasite prevalence are taking place at the same level and rate in both sprayed and non-sprayed villages, i.e., the reductions were adjusted for the background differences. This formula was used only when the denominators were non-zero.

Ethical clearance
Ethical permission for the study was obtained from the Ethiopian Public Health Institute and Amhara Regional Health Bureau. Verbal consent was also obtained from the owner of each house sampled for mosquitoes. The study did not involve human or animal subjects. Table 1 shows the abundance and abdominal status of A. arabiensis collected before and after spray. The abundance and abdominal status of A. arabiensis varied by sampling method, spray status, study village, and year. Among 5425 A. arabiensis, 3111 of them were collected by CDC light traps and 2314 of them by pyrethrum spray catches (PSC). The number of semi-gravid and gravid A. arabiensis was smaller in CDC light trap catches than in PSC collections. The proportions of unfed A. arabiensis were higher in CDC light trap catches than in PSC collections. Fresh fed A. arabiensis was dominant in PSC collections (>75%), while <54% of them were FF in CDC light trap catches. The abundance of these FF mosquitoes was declined after IRS in sprayed villages (n = 62), while the number of FF remained high in non-sprayed villages (n = 1690). The abundance of unfed, gravid, and semi-gravid mosquitoes also decreased after spray.

Effect of IRS on HBI
Among 3451 FF A. arabiensis collected, 1574 (45.61%) of them were tested to determine their blood meal sources and sporozoite infection rate. The relative adjusted reduction in A. arabiensis human blood feeding index (HBI) due to IRS implementation varied from 3 to 10% except in 2014 when no human blood was detected in any of the three mosquitoes that were collected and tested. Despite IRS implementation reduced HBI, a non-negligible proportion of A. arabiensis still fed on humans ( Table 2).

Effect of IRS on SR
The estimated sporozoite rate in A. arabiensis was low in both sprayed and non-sprayed villages especially after IRS implementation. As indicated by ELISA test, P. falciparum was more prevalent than P. vivax in both sprayed and non-sprayed villages. Pv-247 was the only subspecies detected during the study period. There was no any mixed infection in the vector in both study villages during the study period. Neither P. falciparum nor P. vivax was not detected in A. arabiensis collected from sprayed villages after the implementation of

Effect of IRS on EIR
The reduction in EIR after the implementation of IRS had similar trends with the reduction in SR because EIR is the product of SR and HBI.

Discussion
The aim of vector control using IRS and LLIN interventions is to reduce vectors' abundance, survival, contact with human, and feeding frequency [45]. Vector abundance is an important determinant of malaria transmission [18,19], and thus factors that increase or decrease vector abundance could have an impact on the intensity of disease transmission. The present study demonstrated that IRS implementation brought about 4-9% reduction in the abundance of A. arabiensis signifying that the abundance of this vector could not be reduced to non-detectable level by the implementation of IRS. Previous similar studies in Ethiopia are either missing or unavailable to compare and contrast with the present study. However, studies from Zambia [9] validated that the effect of IRS on the density of A. arabiensis was not as strong as on A. gambiae s.s and A. funestus due to its exophilic and wide-ranging feeding behavior. Alegana et al. [3] also confirmed that IRS intervention reduced the density of A. funestus and A. gambiae s.l disproportionally, twice as high on A. funestus compared with A. gambiae s.l. Thus, malaria transmission through the bites of A. arabiensis could not be intercepted entirely by the application of IRS so that the impact of IRS should be complemented by and integrated with other vector control interventions. Blood meal source analyses indicated that A. arabiensis was found to have strong preferences to bovine and other hosts over human hosts. Similar results from other parts of the country were published in previous studies [20,21,29] where A. arabiensis demonstrated strong blood meal preferences of bovine over human hosts. Similar results were also reported from neighboring Eritrea [44] and Kenya [11]. Contrary to zoophilic, strong athrophilic tendency was observed in A. arabiensis in Zambia [15,32,43]. The potential reason for the differences observed in the anthrophilic tendency of A. arabiensis between East and South African countries would be justified by the differences in their ecological setups and the impact of these ecological differences on the ecology and behavior of A. arabiensis populations in these two sub-African regions. The application of IRS in the present study further reduced the anthropophily of the vector signifying that zooprophylaxis could be considered as a potential malaria vector control strategy in areas having similar ecological setups with the present study site. On the contrary, a considerable proportion of A. arabiensis still fed on human hosts  suggesting that zooprophylaxis alone cannot intercept malaria transmission. Thus, zooprophylaxis would advance the effectiveness of malaria interventions if used in an integrated way with other vector control intervention measures. Either data are unavailable or no previous attempts were made about the impact of IRS on SR in Ethiopia. However, studies from other African countries [10,35] demonstrated that the implementation of IRS reduced SR to non-detectable level, which is consistent with the results of the present study. And these would substantiate the contribution of IRS implementation in reducing malaria transmission risks in general and SR in particular in the present study area and others having similar ecological setups.
In the present study, P. falciparum was more prevalent than P. viva in A. arabiensis. No A. arabiensis was found positive for either P. falciparum or P. vivax in sprayed villages after IRS. Although too few A. arabiensis were recorded in sprayed villages after IRS, it would have been necessary to process thousands of mosquitoes to find any of them were infected by malaria parasites. There was no any mixed infection detected. The proportion of plasmodium-infected A. arabiensis was also low in non-sprayed villages indicating that SR might be low in naturally occurring vector population. Contradictory results about the prevalence of P. falciparum and P. vivax in A. arabiensis have been reported from different parts of Ethiopia at different times. Massebo et al. [29] reported the dominance of P. falciparum over P. vivax, while [41] reported the dominance of P. vivax over P. falciparum in South West Ethiopia. Animute et al. [4] reported the dominance of P. vivax over P. falciparum in South Central Ethiopia. Differing from all these, [42] reported that no A. arabiensis was found positive either for P. falciparum or P. vivax in South West Ethiopia. Except Taye and his colleagues [42], other investigators used either CDC or PSC mosquito sampling method so that the differences observed in the prevalence of malaria parasites in A. arabiensis could be potentially justified by the differences in ecological setups of the study sites and time period in which the study was conducted. Otherwise, this would be a question of validation.
Malaria transmission intensity, which is normally expressed by EIR, is highly variable with annual EIRs ranging from < 1 to >1000 infective bites per person per year in Africa [6]. Variations in EIR in malaria vectors could be due to different factors such as ecological heterogeneity at continental, regional, and country level [4,22,33] and season (dry or wet) [4,27,40]. For example, the burden of malaria is high in tropical countries having warm temperature, heavy rainfall, high humidity, and efficient Anopheles vectors than nontropical countries [7]. Previous studies indicated that the impact of wet or dry season on EIR is inconsistent, i.e., published reports indicated that EIR is higher during wet season [13,24,40] or vice versa [23,39].
In the present study, a very high Pf-EIR was observed in the vector in both years and study villages although SR and HBI were low. The trend was also similar for Pv-247 EIR in both study villages before IRS in 2013. These findings would be justified by the occurrences of high mosquito density during the study periods. The level of EIR of both parasites went to zero in sprayed villages after the implementation of IRS suggesting that IRS application is 100% effective to control disease transmission. In contrast, previous studies reported that EIR was 90% lower in the ITN community and 93% lower in the IRS community, relative to the community without intervention. The differences observed between the present and previous studies would be attributed to heterogeneity in the ecology and behavior of the vector.
Variation in EIR could also differ by mosquito collection methods [38]. [25] indicated that PSC might underestimate the HBR, which again underrates EIR. Previous studies also reported CDC light traps were more efficient than PSC to estimate EIR [16,17,29,34]. Contrary to these, a study from Bioko Island, Equatorial Guinea, demonstrated that CDC light traps failed to determine the human-biting rate of the anthropogenic A. gambiae s.s [36]. Different from all previous reports, the present study indicated that higher EIRs were recorded from PSC catches than CDC light trap catches. Both CDC and PSC are reported to have shortcomings in mosquito sampling. While CDC light traps attract fed indoor-resting mosquitoes [26,37], PSC tends to miss mosquitoes that leave the house after feeding including those entering the house after feeding outdoor [31]. Therefore, estimating the HBR using either CDC light trap or PSC has limitations, and the need to develop standard HBR remains high. Thus, the differences observed between the present and previous studies might be associated with limitation stated for each sampling method.

Conclusion
This study was linked with IRS application to assess its effect on EIR and other entomological risk factors for malaria transmission. The results illustrated that IRS was strong enough to reduce mosquito abundance, sporozoite rate, and EIR in areas having similar ecological setup with the present study villages.