Molecular xenomonitoring for post-validation surveillance of lymphatic filariasis in Togo: no evidence for active transmission

Lymphatic filariasis (LF), the second leading infectious cause of disability worldwide, is endemic in 73 countries, an estimated 120 million people are infected with the parasites, and 40 million people suffer from complications. The World Health Organization (WHO) outlines specific steps for endemic countries to achieve and document interruption of transmission of LF [1]. Baseline mapping is conducted to identify LF-endemic areas of the country, followed by at least five rounds of annual mass drug administration (MDA) in endemic areas. Concurrent monitoring and evaluation (M&E) are conducted in these areas under MDA to ensure that targets are met and determine when MDA can be stopped. Transmission assessment surveys (TAS) are undertaken after MDA has been stopped, followed by a five-year post-validation period to verify the absence of resurgence.

Togo is a West African country with a population of 6.7 million, with 32% of the population living below the poverty line [2]. The country is divided into six regions and 40 health districts. District medical teams support more than 650 rural health centres nationwide. Eight of the 40 districts were endemic to LF. Those districts belong to three distinct endemic foci and were co-endemic for LF and onchocerciasis.

The results of these surveillance activities demonstrated that there was no ongoing LF transmission in the country [3, 5].

Togo also conducted a national campaign to distribute insecticide-treated bed nets (ITNs) in 2004 followed by subsequent distribution of long-lasting insecticide-treated nets (LLINs) to achieve universal coverage [6–8]. These campaigns likely significantly changed the profile of the LF vector population in Togo, although entomologic studies are lacking.

Recently, Togo became the first country in sub-Saharan Africa to receive WHO validation of the elimination of LF as a public health problem [1, 9, 10]. Before this announcement, the country began molecular xenomonitoring (MX) for LF [11], to collect more evidence demonstrating the absence of LF transmission. This work, presented here, investigated the absence of LF transmission in mosquito populations as recommended by the WHO, with the hypothesis that transmission has already been interrupted in Togo [12].

Anopheles gambiae (s.l.) (n = 7893), the main vectors of LF in West Africa, was the dominant mosquito species (72.6%) collected using the PSC, followed by Culex quinquefasciatus (n = 2942), Aedes (n = 15), Mansonia (n = 3) and six other mosquito species. The mean number of An. gambiae collected per village using the PSC was 213. For the ETC, 2333 An. gambiae, 1374 C. quinquefasciatus, 47 Aedes and 44 Mansonia were collected. The mean number of An. gambiae collected per village using ETC was 63. For HLC, a total of 656, 233, 12 and 10 mosquitoes were reported for the An. gambiae, C. quinquefasciatus, Aedes, and Mansonia species, respectively. The mean number of An. gambiae collected per village using HLC was 82. Based on HLC, the overall biting rate was at the highest in October 2016, with 23 bites/person/night (bpn) for indoor and 22 bpn for outdoor collections, and lowest in February 2017 with 1 bpn for both indoor and outdoor collections.

In Togo, MDA conducted during the past ten years has dramatically reduced lymphatic filariasis (LF) incidence in all implementation units (IU) as demonstrated by the low prevalence of circulating filariasis antigenemia obtained during the two TAS conducted during the five years after MDA was stopped [3]. Despite the very low infection rate in the human population, and the prior submission to WHO of a dossier of evidence supporting the elimination of LF as a public health problem in Togo, this study of W. bancrofti in the vector was conducted to confirm the results of the epidemiological surveys. As WHO states, the establishment of post-validation surveillance is important to prevent recrudescence of infection and renewed transmission [9]. Efficient and sensitive methods are needed to detect this recrudescence risk, and molecular xenomonitoring is gaining recognition as one of the tools to be employed as a surveillance tool in the endgame [15, 21, 24–27].

The molecular analysis of the An. gambiae in this study revealed no infection with W. bancrofti, which is consistent with the very low antigen prevalence observed during the TAS conducted from 2010 to 2015 [3, 5, 28]. In this study, culicine mosquitoes were not analysed, based on the assumption that they are not known vectors of LF in West Africa [29]. In fact, studies conducted in Togo in the 1960s revealed An. gambiae, An. funestus, An. pharoensis, Mansonia africana and M. uniformis to be naturally infected with W. bancrofti. The authors suggested that only the first three species seemed to be vectors, with An. gambiae being a proven local vector of W. bancrofti in Togo [30]. Nonetheless, more recent surveys in Kano (Nigeria) revealed L3 W. bancrofti infection in Culex quinquefasciatus, with others also reporting infection and infectivity rates in Culex mosquitoes [29, 31, 32]. As such, in the light of changing vector abundance and transmission dynamics, future surveys should also focus on mosquito species previously considered as non-vectors, since parasite DNA can be detected in both vector and non-vector mosquitoes [21, 33].

WHO recommendations for the use of PCR on pools of mosquitoes for end-point assessment and post-MDA surveillance suggest that molecular xenomonitoring and mosquito sampling assessment should focus on individual villages (or a cluster of villages when villages are small), rather than on implementation units [16]. Here, the mosquitoes were processed according to the village and IU of the collection. However, whether processing approximately 200 mosquitoes per village (especially where microfilariae positive individuals were identified during surveillance) is enough to identify residual infection remains a matter of debate. Further statistics and modelling analyses might be required to resolve this challenge. While the sample sizes per IU were met, the possibilities of missing communities with the low residual transmission remain, since not all villages in the IUs were assessed. Implementing the WHO recommendation will require (i) more frequent mosquito collections per month in each community to have a large enough sample size, and (ii) mosquito collection in many more villages. These will ultimately increase the cost of xenomonitoring surveys. Finally, while the mosquitoes processed by pool screening were not presented according to catch type and species, due to the absence of infection, the number of mosquitoes collected per village using PSC was higher. A recent study showed that collecting mosquitoes using PSC may be a better tool for xenomonitoring compared to ETC [34]. As such further studies may focus on using PSC alone or other epidemiologically relevant tools such as gravid traps.

Given the very low parasite detection during TAS, we wanted to collect and analyse as many mosquitoes as possible, to enable the detection of very low W. bancrofti infection in the mosquitoes. While the survey was initially planned for all four EUs in the country, with a large number of sites per EU (for good coverage, considering that this was the first post-validation survey for the country), we ultimately could not achieve both with the available budget, so we elected to conduct the survey in many households but in only the highest risk EU. The cost for the survey covered by the programme was approximately 35,910.00 US$. The major cost components were the personnel (34%), laboratory processing (32.3%), field supplies (15%) and transportation costs (8.6%). It is, however, worth mentioning that additional reagents (~3000.00 US$) were donated, by the Noguchi Centre, to enable processing of as many samples as possible. Based on these, the mean approximated cost spent per village is 1051 US$ (including personnel, and sample processing costs). In designing the sampling strategy, various factors and scenarios were considered to arrive at the most cost-effective design. The preferred sampling strategy will differ from country to country, and therefore so will the cost differ [35].

The post-validation molecular xenomonitoring survey in Togo has revealed the absence of infection in the An. gambiae vectors of LF, confirming the results of TAS and surveillance activities that led to the validation of elimination of LF as a public health problem in Togo. The application of molecular xenomonitoring was not without challenge. However, we have demonstrated that large-scale xenomonitoring is feasible. In addition to the challenges of protocol development, sample sizes for evaluation and availability of technical expertise will need to be addressed in order to enable Togo and other country programmes to integrate molecular xenomonitoring into their pre- and post-validation surveillance activities. A further challenge for Togo is to extend the same survey to the five remaining implementation units in the country. Given the proximity of areas of ongoing LF transmission in neighbouring countries, Togo will need to continue epidemiological and entomological monitoring and promote key prevention strategies, such as ITN use, to preserve its achievement and keep Togo a country where future generations can live free of LF.