Postintervention Immunological and Entomological Survey of Lymphatic Filariasis in the City of Olinda, Brazil, 2015–2016

ABSTRACT.

Lymphatic filariasis (LF) is a leading cause of disability due to infectious disease worldwide. The Recife Metropolitan Region (RMR) is the only remaining focus of LF in Brazil, where the parasite Wuchereria bancrofti is transmitted solely by the mosquito Culex quinquefasciatus. This study reports the results of transmission assessment surveys and molecular xenomonitoring in the city of Olinda, RMR, after nearly 15 years (2015–2016) of interventions for LF elimination. Participants were screened for W. bancrofti antigen via immunochromatographic card tests (ICT) in: 1) door-to-door surveys conducted for all children aged 5–7 years from 4 out of 17 intervention areas treated with at least five annual doses of mass drug administration (MDA), and 2) a two-stage cluster sampling survey of residents aged 5 years and older in non-MDA areas. Mosquitoes were collected via handheld aspirators in four MDA areas, differentiated by species, sex, and physiological status, pooled into groups of up to 10 blood-fed, semigravid, and gravid mosquitoes, and screened for W. bancrofti infection by real-time quantitative polymerase chain reaction (RT-qPCR). All 1,170 children from MDA areas and the entire population sample of 990 residents in non-MDA areas were ICT negative. In MDA areas, a total of 3,152 female Cx. quinquefasciatus mosquitoes in 277 households (range, 0–296 mosquitoes per house) were collected via aspiration. RT-qPCR of 233 pools of mosquitos were negative for W. bancrofti RNA; an independent reference laboratory confirmed these results. These results provide evidence that LF transmission has been halted in this setting.

INTRODUCTION

Lymphatic filariasis (LF) is a vector-borne, neglected tropical disease that is a leading cause of disability due to infectious disease according to the WHO.¹ In 2000, it was estimated that over 120 million people were infected with LF parasites, with 40 million suffering clinical hydrocele and lymphedema.² Currently, it is estimated that 863 million people in 50 countries worldwide are at risk for becoming infected with the causative parasites of LF.³ Lymphatic filariasis transmission may involve one or more species of mosquitoes, principally Culex quinquefasciatus, carrying one or more of three parasites. Wuchereria bancrofti is responsible for nearly 90% of global LF infections.⁴ In the Americas, LF is currently endemic in Brazil, Haiti, Guyana, and Dominican Republic and is transmitted only by mosquitoes of the Culex quinquefasciatus species and caused by the filarial parasite Wuchereria bancrofti. In this region, an estimated 13.4 million people are at risk of infection, of which 90% reside in Haiti.^4,5

In 2000, the WHO launched the Global Program to Eliminate Lymphatic Filariasis (GPELF) in response to the tremendous disability, social stigma, and economic impact caused by LF as well as the availability of effective disease control and prevention strategies.⁶ The GPELF aims to interrupt transmission via mass drug administration (MDA) and integrated vector control/management (IVC/IVM) and manage morbidity of patients already infected.¹ Since GPELF activities commenced, there has been a 74% reduction in the number of people infected and the disease has been eliminated as a public health problem in 17 countries.⁷ A recent modeling study estimates that the number of 199 million LF infections in 2000 decreased to 51 million infections in 2018 worldwide.⁸ Despite GPELF successes, the initial goal of LF elimination by 2020 has shifted to 2030.⁹

To interrupt LF transmission, the WHO recommends annual MDA with 65% population coverage for at least 4–6 years to reduce parasitic microfilariae (mf) in the blood of infected persons.¹⁰ Recommendations for MDA differ by region and cocirculating pathogens but consist of a combination of albendazole with either diethylcarbamazine citrate (DEC) or ivermectin to control the parasite (mf).¹⁰ Integrated vector control and management are recommended as additional measures to halt transmission, but there is no gold standard for vector control, so most settings where LF is endemic deploy IVC/IVM based on local preferences and resources. To monitor if transmission has been interrupted, the WHO recommends transmission assessment surveys (TAS) in populations that have undergone at least five rounds of MDA with coverage rates of at least 65%; LF transmission is considered interrupted when LF antigen prevalence falls < 2% in at least two sequential surveys after MDA has ceased. In TAS, it is recommended that immunochromatographic tests (ICTs) be used to assess LF antigenemia in children aged 6–7 years, preferably those born after MDA initiation, or in students enrolled in the first and second years of elementary school residing in areas undergoing MDA.^10,11

Molecular xenomonitoring (MX) is the use of molecular methods, such as polymerase chain reaction (PCR), to detect pathogen DNA or RNA in the vector as a proxy for infection in the human population. Molecular xenomonitoring is a promising tool for monitoring MDA and IVC/IVM efforts at curtailing LF transmission and achieving LF elimination.^12–16 Molecular xenomonitoring has been used in a variety of LF-endemic settings, and evidence of its utility exists in five out of six WHO regions, namely, the WHO regional offices of Africa/AMRO (Ghana, Sierra Leone, Tanzania), the Americas/AMRO/PAHO (Trinidad and Tobago), Eastern Mediterranean/EMRO (Egypt), Southeast Asia/SEARO (India, Sri Lanka), and Western Pacific/WPRO (American Samoa, French Polynesia, Samoa).^17–26 Several MX protocols have been developed for Cx. quinquefasciatus, but none are systematically accepted or endorsed, unlike with the WHO-recommended TAS. Reasons for this include differing survey designs (e.g., cluster or village sampling, geographic-based methods), a lack of reliable diagnostics (e.g., dissection, PCR), and a lack of consensus around a systematic strategy that incorporates all the necessary surveillance elements (e.g., sampling, tools, molecular or other vector endpoints).²⁷

In the Americas, Brazil was the first country to halt MDA and is currently in the post-MDA surveillance phase.⁷ In Brazil, LF transmission solely involves the mosquito Cx. quinquefasciatus and the nematode W. bancrofti.²⁸ The interventions to eliminate LF in Brazil were implemented between 2003 and 2017 in the cities of Recife, Olinda, and Jaboatão dos Guararapes, all located in the Recife Metropolitan Region (RMR), Pernambuco State.^5,29,30 Importantly, unlike most other LF-endemic countries, the Brazilian Ministry of Health decided to use only single-dose DEC (6 mg/kg) monotherapy for MDA, citing lack of evidence that DEC-albendazole coadministration was more effective than DEC alone.^5,31,32 By 2018, these cities had halted MDA and implemented TAS 2 and TAS 3, during which no infected children were identified.³²

Olinda is the second most densely populated municipality in the RMR and one of the four municipalities considered endemic for LF in Brazil.³⁰ Prior to Olinda’s Secretary of Health convening its local arm of the Brazilian GPELF, population-based pre-MDA surveys in 1999 found mf prevalence ranging from 1.3% to 6.1% and antigenemia as high as 31.7% using thick blood smears and ICT testing, respectively.^33,34

In 2005, the Filariasis Control Program of Olinda (FCPO) commenced MDA; programmatic coverage was extended to > 50% of the city over the following 5 years (2005–2010).³⁵ In neighborhoods with mf prevalence greater or equal to 1%, all residents aged 2 years of age and older were subjected to MDA. In neighborhoods with mf prevalence less than 1%, passive surveillance through thick blood smears and the traditional intervention strategy (selective DEC treatment of detected cases) were maintained.^30,35 In this same period, IVC/IVM were not consistently deployed, but when used, strategies against Cx. quinquefasciatus included use of the larvicide Bacillus sphaericus and selective destruction of breeding sites.³⁶ Between 2005 and 2014, the FCPO distributed approximately 400,000 doses of DEC, with annual treatment coverage greater than 65%.³⁵

Since the 2005 initiation of GPELF/FCPO activities (MDA, IVC/IVM) in Olinda through 2016, the impact of MDA via DEC alone and selective IVC/IVM has not been systematically evaluated in areas that underwent MDA and those that did not. Furthermore, a comprehensive screening of Cx. quinquefasciatus for W. bancrofti infection has not been performed. This paper reports the results of population-based surveys for W. bancrofti antigen via ICTs in the human population and community-based MX for W. bancrofti infection via real-time (RT)-PCR in Cx. quinquefasciatus populations from four neighborhoods that were targeted for the MDA intervention and in the remaining area of the city (14 neighborhoods) that were not subjected to MDA in Olinda, RMR, Brazil. This comprehensive study, over 1 decade after local GPELF activities commenced, aims to provide crucial data on the state of LF transmission in the RMR, Brazil.

MATERIALS AND METHODS

Site characteristics.

Olinda is the second largest city in the RMR, with 393,115 inhabitants. It has an area of 41.3 km² (36.7 km² urbanized) and is divided into 31 neighborhoods.^37,38 Its tropical climate has an average annual temperature of 26°C (78.2°F). The peak dry season is in November (average of 35 mm of rainfall), while the rainy season, usually extending from June to August, peaks in June and July (average of 170 mm of rainfall).³⁹ Approximately 40% of its households have suboptimal sanitation, and only 17.9% of its urban households are located on public roads with formal indicators of urbanization (presence of manholes, sidewalks, paving, and curbs).³⁷

FCPO.

Initially, FCPO activities covered 14 neighborhoods with evidence of LF transmission (mf and/or antigen prevalence ≥ 1% as assessed by thick blood smear [TBS] and immunocromatrografic card test [ICT]) as detected by pre-MDA population-based surveys in 1999.³⁵ In 2011, the FCPO detected LF transmission via passive surveillance in three additional neighborhoods and therefore added them to programmatic activities, targeting a total of 17 of 31 neighborhoods with MDA.⁴⁰ Figure 1 depicts the Olinda neighborhoods that were and were not subjected to MDA.

Figure 1.

Evaluation areas (EAs) in the mass drug administration (MDA) areas and spatial distribution of clusters (census tracts) in the population sample survey in the non-MDA areas of Olinda, Brazil.

Study design and sampling.

Human surveillance.

Lymphatic filariasis transmission was assessed by two distinct cross-sectional designs: 1) community-based surveys of all children aged 5–7 years (born after MDA commenced) living in four neighborhoods (evaluation areas) that underwent at least five rounds of MDA with coverage rates ≥ 65% (“MDA areas”), and 2) cluster sampling survey of residents aged 5 years and older living in neighborhoods that did not undergo MDA (“non-MDA areas”).

The community-based surveys in MDA areas were slightly different than standard TAS methodology, which recommends surveying a sample of children aged 6–7 years from the community or those enrolled in the first year of elementary school (usually children aged 6–10 years). Instead, this study evaluated all children aged 5–7 years living in the selected MDA areas, which would account for children born shortly after the end of the last MDA cycle and would ensure that children from other (nonselected) neighborhoods were not included in the study.

Circulating filarial antigen screening in humans.

This study used ICT (AD-12-ICT, NOW Filariasis; Binax, Inc., Portland ME) to screen for LF infection. The ICT was performed according to the manufacturer’s instructions and read by trained technicians in the field 10 minutes after the blood sample was added to the sample application pad. Visualization of two lines (test and control) was interpreted as a positive result. Any positive ICT, if detected, would be followed up for confirmation via TBS. Areas that did and did not undergo MDA were screened in slightly different ways, as follows.

MDA areas: community-based surveys in children aged 5–7 years.

From the 17 neighborhoods that underwent MDA, at least nine were subjected to five rounds with treatment coverages above 65%.⁴⁰ Of these nine neighborhoods, the four with treatment coverage above 80%, namely, Alto do Sol Nascente, Águas Compridas, Sítio Novo, and Salgadinho, were selected for door-to-door surveys of the entire resident population aged 5–7 years. Each one of these four neighborhoods was defined as an evaluation area (EA) (Figure 1).

Table 1 presents the population size, number of doses administered, and coverage rates of these EAs. The study design and age range (5–7 years instead of 6–7 years as recommended by the WHO) were selected to ensure the inclusion of all children that were: 1) living in these EAs, and 2) born after MDA was initiated (at least 5 years).¹⁰ An important feature of this study is that children were tested for W. bancrofti antigen via ICT in their household settings, because some attended schools outside of the study area (i.e., school-based TAS would not accurately reflect the potential infection, risk, and ongoing transmission of LF based on geography).

Table 1.

Resident and target population, number of MDA doses, and MDA coverage rates in the study areas of Olinda, Brazil, in 2016

EAs	Resident population, n	Target population (aged 5–7 years), n	MDA rounds, n	Doses administered, n	Coverage rates, %
Alto do Sol Nascente	3,399	143	9	31,394	84
Águas Compridas	20,989	882	8	102,344	83
Sítio Novo	5,662	238	8	22,307	89
Salgadinho	10,426	438	8	38,801	88

EAs = evaluation areas; MDA = mass drug administration.

The sample size was calculated based on an expected prevalence of filarial antigenemia of 1%, with a standard error of 0.5% and 95% CI, yielding a sample size of 1,700 children. This population size corresponded to the population in this age group (5–7 years) living in these four selected EAs, according to demographic census data.³⁷

Non-MDA areas: cluster sampling surveys in residents aged 5 years and older.

A two-stage cluster survey was carried out in a random sample of residents aged 5 years or more from the 14 neighborhoods that did not undergo MDA (Figure 1). Population data by city neighborhoods per census tract (CT) were obtained from the 2010 National Demographic Census. The CT is the smallest territorial unit for obtaining population data and represents approximately 250 households or approximately 1,000 inhabitants.⁴¹ In stage I, a systematic sample of 40 clusters (CTs) was randomly selected among the 250 existing clusters in the total number of neighborhoods that did not undergo MDA. The number of clusters to be sampled was determined based on estimates of at least one case of LF infection in 80% of the clusters in the non-MDA areas. Figure 1 shows the spatial distribution of selected clusters in this survey. In stage II, a systematic sample of households was selected from each of the 40 clusters selected in stage I. The number of households within each cluster was determined based on an estimated 250 households per cluster (CT) and 2.86 residents per household in the eligible age groups (5 years or more) for the study,³⁷ resulting in a sample of 16 households per cluster. The final sample comprised 640 households in the non-MDA area. For household selection, is was calculated a sampling interval of 10, and an initial counting point from the list of households was randomly selected (1–10). All residents more than 5 years old living in the selected households were eligible for the study.

The population sample size was determined considering as parameters a prevalence of filarial antigenemia of 30%,³⁴ based on the data from the pre-MDA survey performed in the city, a standard error of 3%, a 95% CI, a design effect of 2, and expected losses of 30% (conservative for standard expected losses in cross-sectional surveys), yielding a population sample size of 1,600.

Mosquito MX surveillance.

The MX study was conducted in the four selected EAs within the MDA area (Figure 1). Sample size estimates were based on previous mosquito collections and published estimates of freshly blood-fed females needed to accurately detect filarial transmission in post-MDA settings.⁴² The mosquitoes were sampled based on previous MX studies in this exact location by using population-based designs appropriate for the MX development targeted for LF elimination.^42,43 Those studies sought to maximize the numbers of Cx. quinquefasciatus mosquitoes caught in a relatively small geographic area, while preferentially selecting those more likely to be infected, as well as to measure the spatial dispersion of Cx. quinquefasciatus mosquitoes. Thus, this collection was not powered to estimate the extent of infection in mosquitoes (e.g., with purposeful sampling methods related to high human infection prevalence, which was not assumed here given the phase that Olinda was in overall LF elimination activities).

The study was designed to use a fixed geospatial grid, with grid spacing based on measured flight ranges for Cx. quinquefasciatus and Aedes aegypti mosquitoes (which previous estimates from this area suggested could be potentially 10% of the total mosquito yield) from this exact study area.⁴² A previous 2015 pilot MX conducted in Sítio Novo, one of the selected EAs, measured an average and maximum Cx. quinquefasciatus flight range of 50 m and 85 m, respectively; however, these pilot experiments occurred in the most population-dense and vector-productive areas of Sítio Novo and therefore were taken as conservative measures compared with the likely average of the four EAs, considering their population structures and environmental conditions.⁴² In this study, flight ranges of Ae. aegypti have been estimated to be 100 m in Brazilian and other global urban areas, and it was of interest to incorporate analyses from their collections as well as those of Cx. quinquefasciatus due to recent arboviral epidemics of chikungunya virus, dengue virus, and Zika virus. Thus, a geospatial grid of 100 m × 100 m was superimposed over the four EAs that underwent MDA to select houses for vector sampling to incorporate the average flight ranges of all potential mosquitoes of interest (Figure 1).

For development of the 100 m × 100 m grid, satellite images and geographic information systems (GIS) software of ArcGIS 10.2 (ESRI 2014 ArcGIS Desktop release 10; Environmental Systems Research Institute [ESRI], Redlands, CA) and QGIS 2.10.1 (QGIS Development Team [2015], QGIS Geographic Information System, Open Source Geospatial Foundation Project; http://qgis.osgeo.org) were used to visualize the study area and select houses for vector sampling. First, ArcGIS was used to create a 100 m × 100 m fixed spatial grid, after which the ArcGIS grid was superimposed over the satellite image of the four EAs as identified via shape files that were imported into ArcGIS/QGIS from the Brazilian national census.³⁷ Next, each individual grid square was reviewed for visualization of a residential structure at or near the centroid of each grid square so that these grid squares could be included in the study.

Superimposing the 100 m × 100 m grid over the four EAs yielded 438 grid squares, of which approximately 30% (129 grid squares) were deemed unsuitable because of the lack of residences—usually due to mangrove, riverside, forest, or otherwise unoccupied areas—but these areas were distributed equally throughout the four neighborhoods, thereby conserving the integrity of selecting households for mosquito aspiration in relation to population size and territory.

A total of 309 grid squares were selected for aspiration over the four neighborhoods. To address potential environmental factors related to mosquitoes (e.g., differences in abundance due to altitude), the grid incorporated two neighborhoods located at or near sea level (Sítio Novo and Salgadinho) and two neighborhoods in hills of 35 m (Águas Compridas) and 70 m (Alto do Sol Nascente).⁴⁴

Data collection.

Transmission assessment surveys.

MDA areas.

Community-based ICT testing occurred between October 2015 and October 2016. Using maps and lists of existing households provided by the FCPO, community health workers initially visited the houses to identify and register those with eligible children in the study age range. Next, study field teams visited the houses to inform the head of household (HoH), parents, or guardians about study objectives and asked for permission for the eligible children to participate. Households were visited at least three times if children and those who could provide consent for them were not available at the time of the visit. Once they agreed to allow the eligible child in their care to participate, they were invited to read and sign the consent form on behalf of the child. Next, the field team collected the child’s individual information (age, sex, prior DEC treatment) and a 100-μL fingertip blood sample to screen for W. bancrofti antigen via ICT.

Non-MDA areas.

Population-based ICT testing occurred between May 2016 and December 2016. During the household visit, individual data of the participants were obtained using the same questionnaire used in the survey carried out in the EAs (MDA areas), and a 100-μL fingertip blood sample was collected to perform the ICT.

Mosquito collection.

Mosquito samples were collected in all selected MDA areas between August 16 and September 15, 2016, coinciding with approximately 1 month after the end of rainy season (peak mosquito abundance). Residences located at the centroid of each populated quadrant were selected as the preferred aspiration point for vector collection. After satellite imagery and ArcGIS/QGIS selection and prior to field deployment, the coordinates of each selected house were preprogrammed into geographic positioning systems (GPS) devices (Garmin GPSmap 76CS, 3-m precision).

Once the HoH agreed verbally to participate, the HoH was interviewed via a structured questionnaire to obtain individual information (age, sex, previous LF diagnosis, and prior history of MDA), while the other study team member aspirated the house for 10 minutes via the aspiration protocol described below. If the HoH declined or the house was not available (e.g., had been razed to the ground including as the result of landslides), alternate houses were selected as follows until a suitable alternative could be enrolled: one house to the left, one house to the right, the house opposite the originally selected house.

The details of the handheld battery-powered aspirator used in this study (Horst Armadilhas [www.horstarmadilhas.com.br]) and further particulars, including pilot development of aspiration protocols, are described elsewhere.⁴² In this study, houses were aspirated for 10 minutes using an electronic timer, with the allocation of collection depending on house structure but via the following protocol that takes 5 minutes of collection in each of the following areas: 1) living room, 2) bedroom(s), and 3) internal toilet or external toilet/septic tank/water storage tank/water distribution box. Houses were aspirated daily from 9:30 am to 12:00 pm and from 1:00 pm to 3:00 pm, with the morning sessions expected to preferentially select nocturnal Cx. quinquefasciatus mosquitoes (after night-time feeding and while still resting on walls indoors in the mornings), whereas the afternoon sessions were expected to preferentially select for day-biting Ae. aegypti mosquitoes. These hours also considered security and logistics, including allowing post-sample collection processing and storage each day.

Each week, the four EAs were placed on a rotating schedule wherein for each day of the week, houses would be selected for aspiration. This allowed the spatial (mosquito collection in each area each week) and temporal (mosquito collection in each area over 4 weeks) comparison for the study period.

Mosquito sample handling.

Prior to field collection, mosquito collection nets were prelabeled with the house identification number, date, and time of collection. After the collection, mosquito nets were placed in an open-top storage box with –20°C freezer blocks and transported back to the Instituto Aggeu Magalhães (IAM)/FIOCRUZ-Pernambuco (PE) Department of Parasitology within 2 hours of field collection. Next, nets of collected mosquitoes were immediately placed flat on shelves in a –20°C freezer for at least 20 minutes for immobilization. Each mosquito net was then removed from the freezer and placed on ice for processing by trained (a team of one undergraduate and one doctoral level) entomologists as follows: 1) separation by species (primarily Cx. quinquefasciatus and Ae. aegypti); 2) separation by sex; 3) determination of physiological status for Cx. quinquefasciatus females as unfed, blood fed, semigravid, and gravid. Finally, female Cx. quinquefasciatus mosquitoes were placed into Eppendorf tubes of mixed physiological status (as were male Cx. quinquefasciatus and all Ae. aegypti mosquitoes, each separately) and stored at –80°C for future analysis. All Eppendorf tubes were labeled by house identification number, date, and time of collection and sorting.

Next, female Cx. quinquefasciatus mosquitoes were removed from –80°C storage and held on ice, and all unfed mosquitoes were sorted into pools of 5–10 females of roughly the following proportions per pool: 20% gravid, 40% semigravid, and 40% blood fed. As mosquitoes were collected from each house only once during the study period, each pool was created to represent an estimate of (potential) W. bancrofti infections per house per day for that individual house.

Laboratory procedures and techniques.

Molecular screening of female Cx. quinquefasciatus mosquitoes.

Total RNA extraction was performed using a RNeasy mini kit (Qiagen, Valencia, CA; catalog number 74106) according to the manufacturer’s specifications. Briefly, each female mosquito pool (up to 10 mosquitoes/pool) was mechanically disrupted and homogenized in a 2-mL round-bottom microcentrifuge tube containing 600 µL of red blood cell lysis buffer with 2-mercaptoethanol and one stainless-steel bead (5-mm mean diameter) using a TissueLyser II (Qiagen) at 26 cycles/second for 3 minutes. The homogenates were then centrifuged (13,000 × g for 10 minutes), and the supernatant was carefully collected and transferred to clean microcentrifuge tubes. All RNA extracted was stored at −80°C before further analysis. RNA samples were reverse transcribed using a Qiagen QuantiTect® reverse transcription kit according to the manufacturer’s instructions. Successful generation of complementary DNA (cDNA) was confirmed by real-time PCR assays targeting the Cx. quinquefasciatus S7 mRNA gene using Qiagen QuantiTect SYBR green master Mix.

Wuchereria bancrofti screening was undertaken using a SYBR green real-time PCR targeting a fragment of the long dispersed repeat (LDR) in the nuclear scaffold/matrix attachment region gene of W. bancrofti (GenBank accession number AY297458). Primers used were Wuch FOR (5′-TTTGATCATCTGGGAACGTTAAT-3′) and Wuch REV (5′-TGATAACCAGAGATCCACCGTA-3). The 66-base-pair target sequence of the LDR gene (5′-TTTGATCATCTGGGAACGTTAATATATCTGCCCATAGAAATAACTACGGTGGATCTCTGGTTATCA-3′) was used to generate a synthetic standard to use as a positive control. Molecular results from each individual PCR were analyzed using the 7500 Fast Software v. 2.0.6 prior to exporting and combining all the data for each PCR test, for each sample, within an Excel database.

A reference laboratory from the CDC, Atlanta, GA, was enlisted for independent confirmation of MX results. Wuchereria bancrofti confirmatory detection was undertaken using a TaqMan real-time assay targeting the same gene region as described above. Primers used were WB-LDR1 FOR (5-ATTTTGATCATCTGGGAACGTTAATA-3′) and WB-LDR2 REV (5′-CGACTGTCTAATCCATTCAGAGTGA-3′) with a hydrolysis probe (5′-FAM- ATCTGCCCATAGAAATAACTACGGTGGATCTCTG-BHQ1-3′, where FAM is 6-carboxyfluorescein and BHQ1 is black hole quencher 1). PCRs were prepared using 12.5 µL of TaqMan quantitative PCR (qPCR) master mix, final concentrations of 0.4 µM each primer and 0.4 µM probe, 9.2 μL of nuclease-free water, and 1 µL of template cDNA for a final reaction volume of 25 µL. Prepared reactions were run on a real-time PCR system for 2 minutes at 50°C and 10 minutes at 95°C, followed by 40 cycles of 95°C for 15 seconds and 60°C for 1 minute. The increase in FAM fluorescence was monitored in real time by acquisition during the combined annealing/extension step of each cycle using the ROX™ (ThermoFischer Scientific, Waltham, MA) brand of fluorescent dye as the passive reference dye.

Data analysis.

The sample size calculations, data entry, and analysis of the human surveys were performed using Epi Info™, v. 3.5.2, STATA 14 (Stata Corp., 2015), and Microsoft Excel. All data were double entered to minimize errors, and congruence between entered data was verified at each stage. The frequency distribution of the main characteristics of the study population and settings was described. Frequency distributions (number, median, and range per house) of female mosquitoes were described according to physiological status. For molecular analysis, data from mosquito collection and sorting were compiled in Excel to construct strategies for mosquito pooling. Molecular results from each individual PCR were analyzed using 7500 Fast Software v. 2.0.6 prior to exporting and combining all the data for each PCR test, for each sample, within an Excel database.

The usual methods for calculating CIs for clustered data, based on normal approximations,⁴⁵ are not applicable when all clusters have zero prevalence. Hence, we used the following approach. In each area, the Jeffreys Bayesian prior distribution was used for the prevalence. When all n individuals are negative, the posterior distribution is beta (0.5, n + 0.5).⁴⁶ The overall prevalence was calculated as the average of the area-level prevalences, weighted by the area populations. In general, the sum of beta distributions is not a beta distribution,⁴⁷ so the distribution of the overall prevalence was calculated by Monte Carlo simulation, drawing 100,000 times from the set of area-level posterior distributions. The human population size was taken from the above-mentioned census data. For mosquitoes, the population size was assumed to be proportional to the observed number of mosquitoes per house, multiplied by the same human population. Finally, the upper 95% confidence limit of the prevalence was estimated at the 95th percentile of the Monte Carlo sample. For the MDA areas, the results apply only to the four areas in question. However, for the non-MDA areas, the results are generalizable to all of them, because a cluster sampling scheme was used. This area-level method gives a higher upper confidence limit, relative to treating the data as a single proportion, and would do so even for a self-weighting sample. In this sense, the method is conservative. This is because, for small proportions, the variance of the beta posterior is approximately inversely proportional to the square of the sample size,⁴⁸ rather than the sample size itself, in the case of the binomial.

RESULTS

Human surveillance.

MDA areas.

Of the 15,559 households in the four EAs, 1,263 (8.0%) had children in the target age group (5–7 years old), of which 175 were not included (83 refused, 44 were not found or were not accessible, and 48 were closed after at least three visits), totaling 1,088 households. Among the 1,345 children eligible, 91 refused and 84 were not at home, leaving 1,170 (87.0%) who participated in the survey.

Among children examined, 51.0% were males, had a mean age of 5.9 (± 0.7) years, and had lived in the neighborhood for more than 5 years; none of them had a previous LF (mf or circulating filarial antigen [CFA]) history (Table 2). All 1,170 children tested by ICT were negative for W. bancrofti antigen, with the upper 95% confidence limit being 0.4%.

Table 2.

Population and household characteristics by MDA study site in Olinda, Brazil, in 2016

Characteristics	Areas				Total
	Alto do Sol Nascente (N = 159)	Sítio Novo (N = 68)	Salgadinho (N = 302)	Águas Compridas (N = 641)
Territorial area (km2)	0.25	0.39	2.13	1.59	4.36
Population density (inhabitants/km2)	9,284	13,203	4,454	13,203	–
Geographic feature	Hill	Flat	Flat	Hill	–
Household
Registered, n (%)	157 (100.0)	76 (100.0)	279 (100.0)	751 (100.0)	1,263
Visited, n (%)	134 (85.4)	62 (81.6)	264 (94.7)	628 (83.7)	1,088
Mean residents per household (SD)	5.1 (1.8)	4.6 (1.4)	4.4 (1.4)	4.4 (1.5)	–
Mean children per household (SD)	1.2 (0.4)	1.1 (0.2)	1.1 (0.4)	1.0 (0,1)	–
Individual
Eligibles, n (%)	182 (100.0)	82 (100.0)	317 (100.0)	764 (100.0)	1,345
Examined, n (%)	159 (87.4)	68 (82.9)	302 (95.3)	641 (83.9)	1,170
Mean age (SD)	5.9 (0.8)	5.9 (0.8)	5.9 (0.8)	6.0 (0.8)	–
Male sex (%)	81 (51.0)	24 (35.0)	167 (55.0)	323 (51.0)	–
Average residence time in the area in years (SD)	5.2 (1.5)	5.4 (1.3)	5.3 (1.4)	5.7 (1.2)	–
ICT positives,* n (%)	0 (0.0)	0 (0.0)	0 (0.0)	0 (0.0)	0
Report of previous filariasis (%)	0 (0.0)	0 (0.0)	0 (0.0)	1.0 (0.1)	1.0

ICT = immunochromatographic test; MDA = mass drug administration.

^*< 0.4% (one-sided 95% CI).

Non-MDA areas.

Of 640 randomly selected households, 181 (28.0%) houses were not found because of faulty address details or unoccupied. Therefore, a total of 459 (71.2%) households were eligible for the survey, of which 322 (70.1%) agreed to participate and 137 refused. Table 3 describes select characteristics of the general (reference) and study populations. A comparison of age and sex distributions of survey participants and the general population did not yield statistically significant differences (Table 3). A total of 990 people from the 322 participating households were examined via ICT. Of these, the majority (60.1%) were female, had a mean age of 39 years (range, 5–93 years), and had lived in the neighborhood for a median of 20 years. Of the 990 participants, 10 (1.0%) self-reported a previous LF diagnosis and 17 (1.7%) self-reported previous DEC treatment. All 990 residents tested by ICT were negative for W. bancrofti antigen, with the upper 95% confidence limit of prevalence being 1.4%.

Table 3.

Participant characteristics and prevalence of filarial antigenemia via ICT in non-MDA areas in Olinda, Brazil, in 2016

Characteristics	Resident population,* n (%)	Examined, n (%)
Total	183,240† (100.0)	990 (100.0)
Age (years)
5–14	27,670 (15.1)	136 (13.7)
15–24	31,893 (17.4)	161 (16.3)
25–34	33,422 (18.3)	127 (12.8)
35–44	29,582 (16.1)	151 (15.3)
45–54	25,500 (14.0)	179 (18.1)
55–64	17,951 (9.8)	115 (11.6)
65–74	10,594 (5.7)	79 (8.0)
75–94	6,628 (3.6)	42 (4.2)
Sex
Males	90.155 (45.7)	395 (39.9)
Females	107.114 (54.3)	595 (60.1)
Previous treatment with DEC
Yes	–	17 (1.7)
No	–	973 (98.3)
Already had filariasis
Yes	–	10 (1.0)
No	–	980 (99.0)
Prevalence of filarial antigenemia (ICT)‡
Positive	–	0 (0.0)
Negative	–	990 (100.0)

DEC = diethylcarbamazine citrate; ICT = immunochromatographic test; MDA = mass drug administration.

^*Resident population total according to the IBGE Demographic Census 2010.

^†Total number of the population aged between 5 and 94 years.

^‡< 1.4% (one-sided 95% confidence interval).

Mosquito surveillance: MX of Cx. quinquefasciatus.

A total of 3,152 female Cx. quinquefasciatus mosquitoes were captured in 277 households in the four EAs. Of these, 887 (28.0%) were unfed, 1,238 (39.0%) were blood-fed, 440 (14.0%) were semigravid, and 587 (18.5%) were gravid (Table 4). A total of 233 pools of up to 10 Cx. quinquefasciatus females (per day, per house) per pool were analyzed. Upon RT-PCR, all pools showed no evidence of W. bancrofti infection. These results were retested at and confirmed by the CDC.

Table 4.

Culex quinquefasciatus abundance [median, range] by physiological status, year, neighborhood, altitude, and environment

Parameters	EAs
	Águas Compridas	Alto Sol Nascente	Salgadinho	Sitio Novo
Number of houses, n	144	27	76	30
Analyzed pools, n	122	20	69	22

Mosquito samples: physiological status	Collected, n	Median (range) per house	Collected, n	Median (range) per house	Collected, n	Median (range) per house	Collected, n	Median (range) per house
Unfed	321	1.0 (0–36)	108	1.0 (0–40)	385	2.0 (0–103)	73	1.0 (0–17)
Blood-fed	688	2.0 (0–35)	97	1.0 (0–23)	320	1.0 (0–61)	133	1.0 (0–41)
Semigravid	225	0.0 (0–20)	24	0.0 (0–7)	155	1.0 (0–40)	36	0.0 (0–8)
Gravid	290	1.0 (0–30)	41	1.0 (0–9)	208	1.0 (0–27)	48	1.0 (0–15)
Total	1,524	4.5 (0–92)	270	3.0 (0–79)	1,068	7.0 (0–231)	290	4.0 (0–75)

EAs = evaluation areas.

DISCUSSION

This comprehensive assessment of human and vector populations indicates that Brazilian GPELF activities may have effectively curtailed LF transmission in Olinda, RMR, Brazil, to GPELF-recommended levels of < 2% antigenemia in the areas that underwent MDA and IVC/IVM. Human antigen for W. bancrofti as screened via ICT indicated no incident infection in the study area. Molecular xenomonitoring for mosquito infection with W. bancrofti as screened via RT-PCR suggested no infection or infection at levels undetectable by the diagnostic tests used. Strengths of the study include the study design of large population-based studies that considered various factors commonly found to be associated with W. bancrofti infection in human (e.g., socioeconomic) and mosquito (e.g., environment, altitude, sanitation) populations. Study results indicate that the emergence of new transmission in non-MDA areas as well as MDA areas is, for the time being, unlikely.

This study suggests that LF may have been eliminated in Olinda, the second largest urban center in the RMR and one of the only remaining LF transmission sites in Brazil, after five rounds of annual MDA with DEC alone in areas with baseline prevalence of 1–6%. The fact that no infection was detected in either human or mosquito populations provides reasonable evidence that LF transmission may have been interrupted in this area, demonstrating the effectiveness of FCPO activities. Furthermore, it fortifies WHO and GPELF success in the Americas, where MDA has been decreasing since the inception of GPELF in 2000. Among the four countries in the Americas where LF is endemic, Brazil was the first country to discontinue MDA and is currently in the post-TAS surveillance phase to confirm LF elimination.⁴⁰ The results of this study indicate this is the correct posture for Olinda, perhaps the RMR, and maybe even Brazil as a whole.

However, despite the results of this study conducted in 2015–2016 and occurring less than 1 km from the outer limits of the Olinda study area, a cluster of 3 mf-positive human cases were detected in Recife, RMR, in 2017 in an area that had been treated with five rounds of MDA (2008–2012) (Health Department of Recife, unpublished data). This is of great concern, as the RMR has seemingly been on the brink of achieving elimination several times over the last 3 decades but keeps experiencing recrudescence. A recent survey of LF transmission in schoolchildren from nine municipalities within the RMR provided no evidence of transmission in these areas.⁴⁹ Still, and despite the results of the study reported here from Olinda, RMR, it is absolutely critical that further studies comprehensively assess whether LF elimination has truly been achieved in the greater RMR to align with the GPELF goal of total LF elimination in the coming decade.

Human surveillance results compared with GPELF-recommended TAS.

The WHO recommends that the effectiveness of LF elimination strategies be assessed in sentinel areas through TAS after five effective rounds of MDA.¹⁰ In line with WHO recommendations, this study investigated children living in four EAs and born after MDA commenced to assess whether LF transmission may have truly occurred after MDA began. This study also used ICTs, so LF infection screening in the human population was based on a highly sensitive point-of-care test and allowed for reliable data on transmission in the locations tested.

In contrast to the WHO, which recommends school-based surveys in first and second-year primary schoolchildren, this study used door-to-door community-based (as opposed to school-based) surveys over a carefully defined territorial base; this was crucial because in the study setting, many children in this age group attend schools outside their neighborhood of residence. This strategy made it possible to provide more accurate data on LF transmission after MDA.

None of the children aged 5–7 years living in the EAs were positive for W. bancrofti antigen by ICT, providing evidence of the effectiveness of MDA to eliminate LF. Although the parameters used to calculate the sample size in the MDA and non-MDA areas differed from each other, especially in relation to the prevalence parameter, both surveys found zero prevalence, with 95% confidence limits below the 2% threshold for the prevalence of CFA by ICT, which is considered evidence of interruption of LF transmission.¹⁰ These results were confirmed by MX of Cx. quinquefasciatus, in which there was also no evidence of filarial infection in the large sample of insects collected. These results may be related to high MDA coverage (greater than 65%) in Olinda, and microfilaremia positivity indexes lower than 1% in sentinel sites, since FCPO inception.⁴⁰ Similar results were reported by Ramaiah et al.⁵⁰ while assessing LF transmission in areas in India where LF is endemic, where antigenemia was not detected in children aged 0–5 years after 10 annual rounds of MDA using DEC alone, leading the authors to conclude that LF transmission was interrupted after five doses of MDA. Other studies that evaluated the impact of MDA with DEC alone in other countries also demonstrated the effectiveness of this scheme in reducing the prevalence of microfilaremia, parasite load, and antigenemia, reinforcing the evidence of the effectiveness of this scheme in eliminating LF.^51–53

Unlike most endemic countries that follow the WHO recommendation of a treatment regimen of a combination of antifilarial drugs in the GPELF,¹ the Ministry of Health of Brazil recommended MDA with DEC alone in single annual doses of 6 mg/kg.⁵⁴ Brazil’s decision to use DEC alone was reinforced in a recent systematic review that included 13 RCTs and cluster RCTs which concluded that the use of DEC alone and DEC associated with albendazole in MDA had no effect on the prevalence of microfilaremia and antigenemia in the countries that made this comparison.⁵⁵ However, despite the proven effectiveness of MDA with DEC alone for interrupting LF transmission, studies have shown the need for a greater number of doses than the number required for drug combination schemes, generally DEC plus albendazole, mainly in areas with high levels of pre-MDA prevalence.^50,56–61

Molecular xenomonitoring in this study and in comparison with historical and contemporary MX programs.

As LF programs move towards elimination, it is crucial to have appropriate, sensitive, and cost-effective surveillance systems in place to measure whether programs have met their target elimination thresholds, which in turn allow them to sustain success and critically guard against recrudescence. As human infection prevalence declines, the number of humans needed for pathogen screening increases, but active case surveillance is often no longer cost-effective for reasons including that it is difficult to find humans willing to be screened when the common community perception is that LF has already been eliminated. In such situations, vector-based pathogen surveillance via MX may serve as a useful alternative to monitoring human infection and thereby residual transmission.²⁷

Many national LF elimination programs have successfully reduced LF prevalence by MDA and IVC/IVM, after which they have used MX as a complement to TAS for continued surveillance.²⁷ Prior to this study, there have been a total of 18 noncontrolled studies and trials that tested the effectiveness or added utility of MX for LF surveillance and 13 noncontrolled studies and trials that included testing the effectiveness of MDA and included MX as an added surveillance tool. Molecular xenomonitoring and TAS have been assessed in five of six WHO regions. However, there is no universally recommended, standardized method for using MX as a tool to monitor LF transmission in intervention areas.

This study developed and used a new way to sample mosquitoes for MX by using a fixed 100 m × 100 m geospatial grid from which to sample mosquitoes from the centroid of each grid square across the same space and during the same time period where human surveillance was occurring. In terms of adequate sample size, estimates depend on previously existing literature on vector abundance and infection prevalence, and data on this field site were limited, so this study assumed to have low (to no) population infection. Previous protocols for analyzing collections of resting, blood-fed Cx. quinquefasciatus mosquitoes to determine accurate filarial prevalence rates in post-MDA settings have estimated needing up to 7,500 freshly blood-fed females.²⁶ The present study collected only 1,238 freshly blood-fed females and a total of 2,265 females that had previously blood-fed (gravid, semigravid). It should be noted, however, that collecting higher numbers does not guarantee the ability to detect infection, and in the case of this study, a cost-benefit analysis indicated that resources simply could not be stretched to collect greater numbers of mosquitoes.

An earlier study in the same areas yielded over 10,000 Cx. quinquefasciatus females, of which nearly 5,000 were blood-fed, when collections occurred from roughly 9:00 to 11:30 am each day from July to August.⁴² In this existing study, mosquitoes were collected during the first, unprecedented Zika epidemic and then during subsequent chikungunya epidemics in Brazil, and given the strong desire to better understand Ae. aegypti and Cx. quinquefasciatus transmission dynamics, this study team shifted daily collection times to maximize collections across the grid by conducting collections in the morning (favoring blood-fed, indoor, resting Cx. quinquefasciatus mosquitoes) and afternoon (favoring day-biting, mobile Ae. aegypti mosquitoes).

Further, in terms of the time of year of collection, this study collected mosquitoes through August to September, thereby collecting mosquitoes 1 month beyond the peak rainy season. It was always expected that the shift in time of day may result in a greater proportion of Ae. aegypti collections at the expense of that of Cx. quinquefasciatus, but a shift in time of year (moving forward by 1 month) that could possibly result in significantly lower collections was not expected. However, this study collected significantly fewer Cx. quinquefasciatus mosquitoes overall, although only slightly fewer blood-fed mosquitoes proportionally to the earlier collections. It is unclear if the diminished mosquito yields provided a sufficient sample size to detect LF infection (if existent) in what appears to be an area of very low to no corresponding human infection. At a minimum, however, this study ensured that mosquitoes were collected across the entire MX grid with each quadrant represented in collections, per the design to provide a geography-based MX system anchoring vector metrics to human collections in the same time and space.

Despite best efforts, this study had some limitations. First, there was a higher-than-expected degree of attrition, largely related to the inability to recruit eligible participants in the area where LF was nonendemic. This was due to faulty data (e.g., address) provided by the FCPO to the study team, house loss (e.g., landslides) or abandonment, and inability to encounter residents despite at least three attempts to do so. Although attrition was above 20% in the population-based study in the non-MDA area, a comparison of age and sex distributions of the survey participants and the general population did not indicate statistically significant differences, suggesting that selection bias may not have been a large issue. Similarly, it is unclear if the adjustments in mosquito collections (time of day, time of year) may have negatively impacted the ability to detect LF infection in mosquitoes if it was indeed present.

Recommendations for postintervention surveillance to prevent recrudescence.

Several studies have reported LF transmission in previously treated areas, and this problem has been attributed mainly to the mobility of populations from areas of endemicity to areas where LF is under control or to areas where LF is nonendemic but where the vector is present.^50,62,63 Environmental conditions that favor Cx. quinquefasciatus present risk for LF transmission in areas with high population mobility, which can occur from areas of endemicity to those free of LF, even if such areas appeared to interrupt LF transmission.⁶⁴ Likewise, new LF foci can arise in areas previously subjected to MDA if a considerable portion of the population was not covered by MDA or refused to participate, thereby potentially serving as potential reservoirs to maintain transmission leading to LF recrudescence.^65,66 In recent years, LF-infected individuals from other countries in the Americas where LF is endemic have immigrated to several states in Brazil, and this may represent a risk of reintroduction of the infection in areas with favorable environmental conditions for the reproduction of Cx. quinquefasciatus, the vector of LF in Brazil.^67,68

Thus, once interruption of LF transmission has been verified, routine passive and active surveillance via sensitive diagnostic tools is critical for early detection of any new LF foci. This is particularly the case in settings such as the city of Olinda, where a considerable part of the population lives in poverty, and in areas with precarious urban infrastructure or lack of sanitation and that are adjacent to municipalities where LF is transmitted. Given this and the past experience of LF elimination being almost achieved only to have the disease revert and recrudesce, permanent active surveillance that includes identification and tracking of infection in suspected LF cases is absolutely essential.

Many countries have achieved success in interrupting LF transmission as evaluated through TAS since initiation of the GPELF.^69–72 Considering that CFA tests have limitations when used in areas of low prevalence, there is a need for the complementary use of standardized diagnostic tools for countries that are under surveillance which can detect a recent filarial exposure.^73,74 Therefore, some authors suggest the use of filarial antibody research in long-term post-MDA surveillance.^75,76 A study carried out in post-TAS surveillance areas, with evidence of transmission interruption in American Samoa, Oceania, showed the presence of active LF transmission through these complementary tools.⁷⁷ Screening for filarial infection via antigen and antibody research in the adult population residing in areas under post-MDA surveillance is also recommended, because of the possibility of permanence of residual foci of transmission in this population.⁷⁵

This study in the second largest city of the RMR suggests that MDA and IVC/IVM have curtailed LF prevalence below the WHO-mandated threshold of <2% antigenemia, but at minimum a follow-up survey would be required before LF could be considered locally eliminated. However, a recent cluster of LF cases was detected through parasitological tests in Recife, a city contiguous to the study area by less than 1 km. This provides a stark warning for the RMR, which has nearly eliminated LF but yet has experienced recrudescence more than three times in the last 3 decades, especially in light of the global charge towards LF elimination as led by the GPELF.

In Olinda, the last step of assessing LF transmission by TAS,⁷⁸ among the three steps recommended by the WHO,¹⁰ was completed in 2020, and no infected schoolchildren were detected, confirming the evidence of transmission interruption⁷⁹ in this city. However, and more troubling for the global LF community, this is yet the latest study to demonstrate differing results when human (via parasitological and immunological methods) and mosquito (via PCR) populations are monitored, as is seen in two contiguous RMR cities, Olinda and Recife. It is critical that the RMR and other global settings with such disparate results have open dialogue about how to best approach and resolve these issues, as these and other recent results indicate that WHO-mandated TAS may not (certainly alone) provide sufficient benchmarks on when GPELF members may stop MDA and IVC/IVM.