Prevalence and factors associated with human Taenia solium taeniosis and cysticercosis in twelve remote villages of Ranomafana rainforest, Madagascar

Background Infections with the tapeworm Taenia solium (taeniosis and cysticercosis) are Neglected Tropical Diseases (NTD) highly endemic in Madagascar. These infections are however underdiagnosed, underreported and their burden at the community level remains unknown especially in rural remote settings. This study aims at assessing the prevalence of T. solium infections and associated risk factors in twelve remote villages surrounding Ranomafana National Park (RNP), Ifanadiana District, Madagascar. Methodology A community based cross-sectional survey was conducted in June 2016. Stool and serum samples were collected from participants. Tapeworm carriers were identified by stool examination. Taenia species and T. solium genotypes were characterised by PCR and sequencing of the mitochondrial cytochrome c oxidase subunit 1 (cox1) gene. Detection of specific anti-cysticercal antibodies (IgG) or circulating cysticercal antigens was performed by ELISA or EITB/Western blot assays. Principal findings Of the 459 participants with paired stool and blood samples included ten participants from seven distinct villages harbored Taenia spp. eggs in their stools samples DNA sequencing of the cox1 gene revealed a majority of T. solium Asian genotype (9/10) carriage. The overall seroprevalences of anti-cysticercal IgGs detected by ELISA and EITB were quite similar (27.5% and 29.8% respectively). A prevalence rate of 12.4% of circulating cysticercal antigens was observed reflecting cysticercosis with viable cysts. Open defecation (Odds Ratio, OR = 1.5, 95% CI: 1.0–2.3) and promiscuity with households of more than 4 people (OR = 1.9, 95% CI: 1.1–3.1) seem to be the main risk factors associated with anticysticercal antibodies detection. Being over 15 years of age would be a risk factor associated with an active cysticercosis (OR = 1.6, 95% CI: 1.0–2.7). Females (OR = 0.5, 95% CI: 0.3–0.9) and use of river as house water source (OR = 0.3, 95% CI: 0.1–1.5) were less likely to have cysticercosis with viable cysts. Conclusions/Significance This study indicates a high exposure of the investigated population to T. solium infections with a high prevalence of cysticercosis with viable cysts. These data can be useful to strengthen public health interventions in these remote settings.

Introduction Taenia solium taeniosis/cysticercosis are neglected parasitic tropical diseases mainly affecting people living in Central and South America, South-East Asia, Indian Subcontinent, and Sub-genotypes identified during this study are available from the GenBank database with the following identifiers and accession number (AC): IFAI_Mangevo_051, AC: MT947371 IFAI_Kianjanomby_332, AC: MT947372 IFAI_Kianjanomby_365, AC: MT947373 IFAI_Ampitambe_378, AC: MT947374 IFAI_Ankazotsara_452, AC: MT947375 IFAI_Torotosy_564, AC: MT947376 IFAI_Torotosy_573, AC: MT947377 IFAI_Ampitavanana_641, AC: MT947378 IFAI_Sahavoemba_182, AC: MT947379 IFAI_Sahavoemba_190, AC: MT947380 2-All relevant data are within the manuscript and its Supporting Information files 3-Some additional data cannot be shared publicly because of confidentiality; it studied specific area and targeted generally all the people leaving in each village. All villages studied are very remote and very isolated whose population does not move and can be easily identified even through anonymous identifiers. Data are available from the Institut Pasteur de Madagascar Data Access / Ethics Committee (contact: ipm@pasteur.mg) for researchers who meet the criteria for access to confidential data.
Saharan Africa [1]. Also referred to as poverty-related diseases, these infections are associated with poor sanitary and hygiene conditions, open defecation, free-range pig-husbandry and lack of meat inspection [2]. Despite having been declared potentially eradicable, T. solium infections remain a serious health and economic problem affecting around 50 million people worldwide [3]. T. solium was identified as a leading cause of death from food-borne diseases, accounting for 2.8 million disability-adjusted life-years (DALYs) annually [4].
T. solium taeniosis and cysticercosis are zoonosis with a biological cycle maintained in the environment between humans and pigs [5]. In the natural life cycle of the parasite, humans are the only definitive hosts developing the intestinal adult form (causing taeniosis). Cysticercosis in pigs and in human (respectively the natural and the accidental intermediary host) is caused by the metacestode larval stage of T. solium (called cysticerci). Humans can therefore develop cysticercosis in muscles, eyes or brain. Neurocysticercosis (NCC) is the most severe presentation of this pathology and the most important parasitic disease affecting the central nervous system [6][7][8][9]. NCC may count for almost one-third of seizure disorders with an estimated 2 million people affected [10,11].
A few community-based studies were carried out in Africa to estimate the prevalence of T. solium infections using different diagnostic tools [12][13][14]. Human cysticercosis prevalence varied greatly across African countries, 0.68% to 21.63% of analyzed samples were found positive for circulating antigens while the seropositivity associated with T. solium antibodies ranged from 7.6% to 34.5%. The prevalence of taeniosis reported in Africa was between 0.37 and 13.8% [12]. Nevertheless, there are still little data available regarding the prevalence of T. solium infections and their associated risk factors for the African Continent especially in rural areas [12,[15][16][17].
In Madagascar, the proportion of the population living below the international poverty line was estimated at 75% in 2019 [18]. Around 35% of the population do not have access to basic toilets [19] and 52% of the population (only 35% in rural areas) have access to drinking water [20]. Agriculture, mainly mixed with livestock, is still very dominant with more than seven out of ten employed individuals. In rural areas, about two-thirds of households are traditional farmers where pig breeding (mainly raised in free roaming) holds the third place [18]. Given the persistence of factors determining the spread of T. solium infections such as traditional pig farming, open defecation and poor meat hygiene requirements, Madagascar is among the most infested countries in the world.
Madagascar is a vast island divided into 114 Districts where taeniosis and cysticercosis epidemiological data are scarce [21]. However, the few data available underline the significant burden of T. solium infections in the Malagasy population. At the national level, the only and most recent sources of data on taeniosis were provided by mapping surveys (Based on Kato-Katz tests) performed through integrated national program of the Ministry of Public Health targeting some Neglected Tropical Diseases (i.e. lymphatic filariasis, schistosomiasis and soiltransmitted helminthiasis). These surveys conducted between 2011 and 2015 identified 54/114 co-endemic Districts for schistosomiasis and taeniosis of which 7 districts were endemic only to T. solium [22]. In 2015, an overall prevalence of 0.7% (ranging from 0.3% to 7.3%) was reported affecting 14 of the 24 Districts surveyed. Regarding cysticercosis, very little data are available in the literature and the data collected by the Ministry of Health are mainly based on clinical suspicions rarely confirmed with neuro-imaging or serological diagnosis (detection of T. solium cysticercal antigens by ELISA and/or anti-cysticercal antibodies by ELISA and/or Western blot/ElectroImmunoTransferBlot-EITB). Indeed, since the existence of human cysticercosis in Madagascar which was first documented by Andrianjafy [23], subsequent data have shown that approx. 20-25% of late-onset epilepsy were related to cysticercosis [24,25]. The seroprevalence of the human cysticercosis in the island ranged from 7% in Coastal regions (Mahajanga and Toamasina) to 14-21% in Central Highlands (Ambositra, Ihosy, Mahasolo and Antsirabe) following the geographical importance of pig farming and pig cysticercosis [26][27][28][29][30][31]. These studies are unlikely to reflect the true situation as they did not cover the whole country and the remote rural areas.
The Ifanadiana District (13 townships) is one of the rural Districts of Madagascar heavily burdened by poverty and reported to have a high rate of intestinal parasites [32][33][34]. In 2015 however, the national mapping program reported 0% of taeniosis prevalence in 14 fokontany (the smallest administrative subdivision of Madagascar) including 21 rural villages (comprising one of the villages targeted by this study) [35]. The main objective of the current study was to determine by community based cross-sectional survey the prevalence of human Taenia solium taeniosis/cysticercosisis and the risk factors associated in 12 remote villages of Ranomafana and Kelilalina townships (Ifanadiana District, Madagascar).

Ethics statement
The study was reviewed and approved by the Ethics Committee for Biomedical Research at the Ministry of Public Health of Madagascar (N˚142-MSANP/CE), and by the Institutional Review Board (IRB) of the Stony Brook University (IRB#874952). The field study was conducted under the supervision of the medical team of ValBio Centre and followed ethical principles according to the Helsinki Declaration. All participants received an explanation of the study goals and procedures prior to enrolment. Written informed consent was obtained from adult participants or parent/legal guardians for the children. Participants detected positive for taeniosis were treated by local medical officers with praziquantel at 10 mg/kg in one dose according to the national guidelines. Before this treatment, in order to recover the whole worm properly and avoid dissemination of eggs in the environment, all carriers of Taenia were cured with niclosamide (2 g orally single dose) and with oral purgative (lactulose, 2 tablespoons) one day before and 2 hours after niclosamide. During the 3 days following the niclosamide treatment, stool samples, collected in large plastic bags, were obtained from all Teania carriers.

Study design and population
This is a cross-sectional prevalence study. Overall, 164 households were randomly selected using an electronic random number generator on the village census. Household members older than 5 years and living in the study area (> 50% of their time) during the last 3 months before the beginning of the study were invited to participate. The study was conducted in the District of Ifanadiana (Vatovavy-Fitovinany Region) Southeastern Madagascar. The investigation was carried out in two rural townships: Ranomafana (21˚15'36" S and 47˚27'17" E, elevation 613m) and Kelilalina (21˚17'8" S and 47˚33'17" E, elevation 613m). The targeted villages were situated close to the ValBio Centre (CVB), a Stony Brook University Research Centre in the Ranomafana National Park. The study population was enrolled in five Fokontany, namely Vohimarina, Tsaramandroso, Mandrivany, Ampitambe and Kianjanomby. In these fokontany, twelve rural and remote villages (Ambinanindranofotaka, Mangevo, Marojano, Sahavanana, Sahavoemba, Kianjanomby, Mandrivany, Ankazotsara, Ampitambe, Bevoahazo, Torotosy, Ampitavanana) were targeted (Fig 1). These villages were chosen as they constitute the rural villages receiving biannual medical consultations by the medical team of ValBio Centre with a deworming of the entire population.
The villages included in each fokontany and the geographic coordinates (obtained using a handheld Garmin global positioning system, GPS) of the twelve villages investigated are listed in supporting S1 Table. The geo-reference data collected were used to map the study sites using QGIS 2.14.19-Essen software (http://www.qgis.org). All twelve villages, situated around Ranomafana Rainforest Park, are only accessible after 2-7 hours of walking with no roads or paved paths. Investigated villages were found to be traditional, rural, and heavily reliant on subsistence agriculture. Village sanitation was poor, with lack of water infrastructure, common practice of open defecation and free-range pig-husbandry.

Socio-demographic, behavioural and knowledge data collection
A community-based survey was conducted over a two-month period (June to August 2016). Results on the prevalence of soil-transmitted helminthiasis and risk factors associated were presented in another manuscript and are not further discussed here [34].
Socio-demographic features, hygiene practices and behavioural data (water source, use of soap, hand-washing, pork-eating, defecation place), knowledge about worms, some clinical manifestations suggestive of NCC (headache and seizure) and antiparasitic drug-taking were obtained using an individual questionnaire in Malagasy. Heads of households were also interviewed about the size of the household, water and sanitation conditions and pig breeding. All participants were informed on the risks associated with T. solium infections.

Stool sampling and Taenia spp. detection by microscopy
After signing the consent form, stool samples were collected from each participant in a sealed 50 mL cup for parasitological analysis. Approximately 100-250 mg of stools was conserved at

Taenia species identification
Taenia species identification (T. solium, T. saginata and T. asiatica) from stool samples of tapeworm carriers was performed by sequencing the mitochondrial cytochrome C oxidase subunit 1 (cox1) gene following PCR amplification. Copro-DNAs were extracted from approx. In a first step, primer pairs described by Yamasaki et al.
[39] were used to perform Taenia spp. specific PCR. However, PCR amplification performed with these primers used either in multiplex or simplex assays showed slight cross-amplifications: T. asiatica primers amplified both T. saginata and T. solium DNAs and T. solium Asian genotype primers amplified T. asiatica DNA. Therefore, common primer pairs able to amplify the three Taenia species were designed, and PCR product were directly sequenced. PCR reactions using these common primers were performed in a total reaction volume of 20 μL containing 0.5 μM of each primer (Sigma Aldrich, Germany), 0.05U of Phusion HF DNA polymerase (New England Biolabs, M0530S), 1X Phusion HF Buffer including 200 μM dNTPs (New England Biolabs, B0518S) and 5 μl of genomic DNA template. PCR cycling condition was carried out according to the following program: initial denaturation step at 98˚C, 30 sec followed by 35 amplification cycles (denaturation at 98˚C, 10 sec; annealing at 58˚C, 30 sec and elongation at 72˚C, 30 sec) and a final elongation step at 72˚C, 10min. All PCR reactions were run on GeneAmp PCR System 9700 Applied Biosystem thermal cycler. Parasitic DNA extracted and purified from T. solium cysticerci isolated from pig was used as intra-run positive control and distilled water as negative control.

T. solium cox1 gene sequencing
The amplicons obtained after cox1 gene amplification by PCR (627 bp) were sequenced in both directions by Genewiz (France). Sequence chromatograms were analyzed using BioEdit software. By removing the beginnings and ends of sequences that were illegible after sequencing, a region of 474 base pairs (starting at the position 601 of the completed cox1 gene sequence) for 10 out of 11 Taenia carriers was finally analyzed for Taenia species identification. The sequences were compared with the 627 base pairs reference regions of each species used beforehand to design the primers.

T. solium genotypes identification and phylogenic analysis
For the T. solium genotypes and phylogenic analysis, the ten sequences obtained in this study were also compared with T. solium Asian and African/American genotype sequences available in EMBL/Genbank databases. Their accession numbers and the corresponding native countries are as followed

Blood collection
After signing the consent form, blood sampling was performed for each participant in order to carry out serological diagnosis of T. solium infections according to the flow diagram represented in Fig 2. Finger prick blood samples (300-500 μL) were collected in microtubes (Sarstedt, 20.1344). The serum was obtained after 10 minutes of centrifugation at 12 000 rpm, transported at 4˚C (cool boxes) within 24 hours at the CVB where they were stored at -20˚C. At the end of the survey, all serum samples were transferred and stored to the Institut Pasteur de Madagascar (IPM) at -20˚C until use for the serological tests.

T. solium glycoprotein antigen preparations
An in-house T. solium glycoprotein antigen suspension was prepared for anti-cysticercal antibody (IgG) detection by ELISA and EITB assays. T. solium cysticerci were harvested from infected pork meat obtained in slaughterhouse (Antananarivo) and stored at -80˚C in PBS buffer (DBPS 1X, Sigma Aldrich D8537) until use. Lentil lectin T. solium metacestode glycoproteins (LLGP) antigens were extracted and purified according to the method initially described by Tsang et al., 1989 with slight modifications [44]. All extraction and purification steps have not been changed except for the foamy lipid residues above the pellet that was removed carefully with a spatula without using Freon (CHFCl 2 ) and the urea solubilization of the pellet was not carried out.

Anti-cysticercal antibody (IgG) detection by enzyme-linked immunosorbent assay (Ab-ELISA)
In-house Ab-ELISA assay was performed as previously reported [27]. Briefly, ELISA 96-well microplates (Nunc 3455) were coated with 100 μL/well of previously prepared T. solium glycoprotein antigens extract at 0.1 μg. mL -1 in DPBS 1X and incubated overnight at 4˚C. After three washing steps in washing solution (1X DPBS, 0,05% v/v Tween 20), non-specific absorption was blocked by adding 150 μL/well of blocking buffer (DPBS 1X, 0.05% Tween 20, 1% casein) and plates were incubated for 2h at 37˚C. After three additional washing steps, 100 μL of serum samples, tested in duplicate and diluted 1:200 in blocking solution, were added per wells and incubated at 37˚C for 2 h. Well characterized serum samples of patients from the neuropsychiatry department of Befelatanana Hospital (tested by EITB assay as described below and by neuroimaging) and serum samples from voluntary donors (tested by EITB assay) as positive and negative controls respectively were systematically included in each plate. Plates were then washed five times and incubated at 37˚C for 1h30 with 100 μL/well of Human antibody anti-IgG conjugated with peroxidase (Sigma Aldrich A8419) diluted 1:10 000 in blocking solution. Plates were then washed five times before a 20 min incubation at 37˚C with 100 μL/well of substrate [(4.5 mM OPD, 3.6% H 2 O 2 ) in 0.2 M citric acid buffer, pH 5.5, Sigma Aldrich, Ref. P9029 and 251275]. The enzymatic reaction was stopped by addition of 100μL/well of 2.5N H 2 SO 4 acid. The optical density (OD) was determined at 492 nm using microplate reader (Labsystem multiscan Plus). The negative cut-off value (0.3 OD) was calculated as the mean optical density (OD) of 38 sera (obtained from European healthy subjects living in T. solium infections non-endemic areas, France) plus 3 standard deviations. A previous study done in Madagascar has reported a specificity of 97.4% and a sensitivity of 96.3% of this in-house Ab-ELISA assay [27].

T. solium cysticerci antigens detection by ELISA (Ag-ELISA)
The serum samples were tested for the presence of excretory secretory circulating antigens of the metacestode of T. solium using the commercially available cysticercosis Antigen-ELISA Kit (ApDia Ltd., Turnhout, Belgium) based on the B158/B60 monoclonal antibodies enzymelinked immunosorbent assay (ELISA) [45]. The ELISA was also used and carried out according to the manufacturer's instructions and the OD was read at 450nm. The Ag-index was calculated following the recommended cut-offs: an Ag-index less than 0.8 classified as negative result, an Ag-index greater than or equal to 1.3 considered as positive and values in between classified as doubtful. The reported sensitivity and specificity of the Ag-ELISA for detecting active cysticercosis in humans were 90% (95% CI: [

Enzyme-linked Immunoelectro-Transfer Blot (EITB) assay
The EITB assay was performed as previously described by Tsang et al., 1989 [44]. The specificity and sensitivity of the EITB for detecting exposure to the parasite was reported ranging from 97% to 100% and from 97% to 98% respectively [44,46]. The usual concentration for LLGPantigens in-house EITB assay was about 30ng/μL per mm of membrane. Briefly, each strip membrane was incubated for 2h at 37˚C with human diluted (1:10) sera. The strips were thereafter incubated for 1h at 37˚C with diluted (1:6000) alkaline phosphatase-conjugated goat antihuman IgG (Sigma Aldrich, A1543). Immune reactivity was detected using BCIP/NBT blue substrate solution (Sigma Aldrich, B3804). Among the seven glycoprotein (GP) bands commonly recognized by antibodies from sera of patients with cysticercosis, only the immune reactivity against one or several GP < 50 kDa (i.e. GP42-39, GP24, GP21, GP18, GP14, and GP13) were considered as positive. As a precaution, tests giving a single 50kDa band were not considered as a positive result in this study. Indeed some studies have shown the lack of specificity of a single positive 50kDa band in EITB assay for cysticercosis [47][48][49].

Statistical analysis
Descriptive analyses on the study population were first conducted. Prevalence of Taenia infections was estimated as the number of samples tested positive divided by the total number of participants provided samples. Data were analyzed with STATA 15 software. Due to the lack of the specificity of Ab-ELISA [47, [50][51][52][53], only the prevalence obtained with EITB and Ag-ELISA was used for the association analysis using STATA software. Variables showing more than 25% of missing data were not included into the statistical analysis. A univariate analysis using logistic regression was performed to determine associations between the prevalence and each risk factor. A p value � 0.2 was included in a model (Initial model) for multivariable logistic stepwise regression analysis. A backward deletion method was used to eliminate the factor one at a time considering a p � 0.05 as statistically significant (Final model). A Confidence Interval of 95% (95% CI) and a significance level of 0.05 were performed for the association analysis and the odds ratio calculation.

Participant characteristics
People of all age and both genders, living in the targeted villages and willing to participate to the study were enrolled. A total of 574 willing participants (� 5 years old) from 12 villages belonging to 164 households were included in this study. Stool and blood samples were obtained from 543/574 (92.7%) participants following the exclusion of thirty-one participants with missing basic data (age or gender) and missing biological samples (stool or serum). Seventy-three and eleven participants were further excluded for the analysis respectively due to the insufficient volume of sera to perform all three serological tests used for cysticercosis diagnosis (i.e., Ab-ELISA, ETIB and Ag-ELISA tests) and due to the completion of the epidemiological data (<25%) for the statistical analysis. The final study population consisted of 459 participants surveyed having all epidemiological and serological data available. Fig 2 shows the flowchart describing the enrolment and analysis performed.
Socio-demographic characteristics and health/hygiene conditions of the study population are described in Table 1. Genders were equally represented (235 males and 224 females, sex ratio 1.05). The median age was 23.9 years, ranging from 5 to 82 years old with a high proportion of adults over 15 (56%) and just over a quarter (27.7%) were children between 5 and 10 years old. The households were composed mainly of large families (52.1% of households with 4 to 6 people, median 6.4 people) and the adult population was mostly composed of farmers (54.5%). The highest education was the elementary level (68.6%), of which 38.3% were adults. Thirteen-point five percent of adults � 15 years old never attended school. More than twothirds of participants lived 6 km or more from the main national road (72.1%) and used river stream as house water source (73.2%).
Hygiene conditions were poor: 52.1% of participants did not have access to latrines and practiced open defecation. Surprisingly, a large part of the study population however declared having the opportunity to wash their hands (96.7%), using soap for washing hands (60.1%) and the habit of washing fruits or vegetables (59.3%) before eating/cooking.
More than half (51.6%) reported not having pigs near where they lived and pork consumption was reported by a high proportion (94.8%) of participants principally from district slaughter (64.7%). Most of the participants (78.9%) declared having a history of deworming drugs with a treatment taken at least twice a year. The survey data also showed that 69.3% of the study population was not aware of the concept of disease spreading within households. Nevertheless 60.1% of the people had knowledge about cysticercosis. Questionnaire analysis also

Taenia species and T. solium genotypes
Of the 11 positive stools in microscopy, cox1 PCR product was obtained from extracted copro-DNAs. Their sequencing analysis amplified cox1 DNA fragments and showed that ten belonged to T. solium species and one was an unspecific amplification. Two T. solium genotypes were identified: T. solium Asian genotype and T. solium African-American genotype. The majority of the sequences (9/10) belonged to the Asian genotype and was detected in stool samples from the villages of Ampitavanana, Torotosy, Ankazotsara, Kianjanomby, Mangevo, Ampitavanana and Sahavoemba. Only one cox1 sequence matched with the African-American genotype and was identified from one of the tapeworm carriers living in the village of Sahavoemba (Fig 3). The phylogenic tree analysis performed with cox1 sequences, obtained from human stools, and the cox1 Asian genotypes sequences (obtained from pig cisticerci) deposited in EMBL/ GenBank databases revealed that the majority (8/9) of the sequences obtained in this study are absolutely conserved. These T. solium Asian genotypes clustered together within the corresponding genotype sequences isolated from pig cisticerci collected in Central, South and Western parts of Madagascar [40,41] and from other countries (India, Nepal, Japan and Thailand). All these cox1 sequences belonged to the Asian T. solium genotype, cluster I (Fig 3). This first cluster was less close to the cox1 DNA specimen collected from the eastern part of Madagascar (Toamasina, cluster II) and distant to those found in Indonesia (Asian T. solium genotype Cluster II) and in other countries such as Korea and China (Asian T. solium genotype Cluster IV). The last cox1 T. solium Asian genotype sequence from Sahavoemba formed a cluster apart (Asian T. solium genotype Cluster V).
The only African-American genotype found in this study shared 100% similarity to the sequences already identified in the Southern part of Madagascar (Toliara, from pig cysticerci) and in other African/American countries (Tanzania and Mexico) forming a cluster (African-American T. solium genotype cluster III) which was less close to those found in Ecuador (African-American T. solium genotype cluster II) and more distant to those obtained in Mexico, Cameroon and Brazil (African-American T. solium genotype the Asian T. solium genotype cluster I). The variation of nucleotides and the corresponding amino acids for each T. solium genotype cluster are shown in the supporting S2 Fig.

Exposure to T. solium parasite: seroprevalence of anti-T solium cysticercus antibodies (IgG)
The presence of anti-T. solium IgGs against metacestode glycoproteins in serum was estimated by ELISA (Ab-ELISA) and EITB methods, which given a fairly similar seroprevalence: 27.5% (95% CI: 23.5-31.7) and 29.9% (95% CI: 25.8-34.2) respectively (Table 3). Females seem to be more exposed to T. solium parasite using Ab-ELISA test (Sex ratio: 0.7) while no differential exposure was observed using the EITB assay (Sex ratio: 1.01). The average age of Ab-ELISA seropositive individuals was 22 years, (ranging from 5 to 79 years old) whereas the average age of EITB seropositive individuals was 24.4 years (ranging from 5 and 74 years old). Using both methods, anticysticercal IgG seroprevalence was quite similar between the three defined age-groups. Depending on the village investigated, the exposure rate found by Ab-ELISA and IgG Western Blot assays ranged between 13% -50% and between 16% -70% respectively (Table 3). At household level, the overall seroprevalence of IgG detected by Ab-ELISA and EITB within the same household in which a tapeworm carrier has been detected was 46% and 34% including 5/11 and 3/11 Taenia carriers respectively with co-infection (taeniosis and cysticercosis).

Prevalence of current cysticercosis infection with viable cysts
Current cysticercosis was detected by measuring the excretory/secretory circulating T. solium antigens based on B158/B60 enzyme-linked immunosorbent assay (Ag-ELISA) [45]. Circulating antigens were detected in 57/459 participants giving an overall seroprevalence of 12.4% (95% CI: 9.7-15.8). Cysticercosis affected all age classes, ranging from 5 to 74 years old with an average age

Phylogenic tree showing the Asian and the African-American Taenia solium genotypes of the specimens obtained in this study and in databases.
Cox1 sequences obtained in this study (from human stool with detected Taenia eggs) are underlined in blue. The sequences obtained from EMBL/Genbank databases (from cysticerci of pig) are indicated with their accession numbers followed by the name of each District/township for genes described in Madagascar and by "FOR" for specimens collected in foreign country. The percentage (%) of bootstrap replicates for the reliability tests of the associated taxa clustered together is shown next to the branch (up and down for the genomic and coding forms respectively).
https://doi.org/10.1371/journal.pntd.0010265.g003 Table 3 (Table 3). T. solium antigen seroprevalences were higher in males (15.3%, sex ratio: 1.7) and in adults over 15 years old (15.2%) but did not reach statistical significance (p = 0.07 and p = 0.32 respectively). The seroprevalence of T. solium antigens varied between 3% and 21% according to the village investigated. Doubtful cases (n = 30, 6.5%, 95% CI: 4.6-9.2), with an Agindex < 0.8, were also detected in the study population. These doubtful cases were higher in men (7.7%, sex ratio: 1.5) and quite similar between the 3 age groups. The rate of these cases, which should be confirmed by a second blood sampling, varied from 0% to 20% according to the village investigated (Table 3). Considering the participants within the same household, the overall prevalence of circulating T. solium antigens was 20% comprising 3/11 Taenia carriers.

Risk factors associated with T. solium infections: univariate analysis
Several epidemiological and behavioral factors were evaluated for their association with T. solium infections and summarized in Table 4. For both serological techniques used (EITB and Ag-ELISA), sixteen variables were firstly assessed using univariate analysis. Univariate analysis suggests that a high number of householders (� 4 people, OR = 2.0, p = 0.001) and open defection behaviors (OR = 1.6, p = 0.02) could be risk factors associated with the presence of T. solium anti-cysticercal antibodies (EITB). With p value � 0.2, housing more than 6 km from the national main road (OR = 1.5), pork consummation (OR = 3.1), district slaughter as meat (pork) source (OR = 0.6) and frequent headache (OR = 0.6) were also included in the multivariate analysis for EITB test (Table 4). Contrariwise, regarding T. solium circulating antigen positivity (Ag-ELISA seropositivity), being male (OR = 0.6 for female, p<0.001) and no use of latrine (OR = 1.6, p = 0.049) could be associated with a higher risk of developing active cysticercosis. Multivariable analysis was completed for Ag-ELISA test with the following factors (with p value � 0.2): adulthood (over 15 years, OR = 1.7), use of stand pipe for house water source (OR = 0.3 for stream river), use of soap for washing hands (OR = 1.5), and no habit of washing veggies (OR = 0.7 for habit of washing veggies).

Risk factors associated with T. solium infections: multivariate analysis
In the multiple regression analysis, the risk factors that were significantly associated with anticysticercal Antibodies (EITB) or cysticercal Antigens (Ag-ELISA) positivity (p-value � 0.05)

Discussion
There is little information available on taeniosis/cysticercosis epidemiology in Madagascar [21]. To the best of our knowledge, the current study is the first community-based survey reporting the prevalence of human T. solium infections using three different diagnostic tests for cysticercosis and analyzing the associated risk factors in rural and remotely region of Madagascar. Little data is available regarding the prevalence of adult tapeworm carriers (taeniosis) who are the source of cysticercosis in humans and pigs [54]. The main difficulty for studying taeniosis especially in Africa is the lack of a simple, sensitive and T. solium specific diagnostic tool. Stool microscopy techniques classically used are not able to discriminate eggs of T. solium and T. saginata and have poor sensitivity and specificity [55]. In Madagascar, national surveys did not report any tapeworm carriers in the district of Ifanadiana [35]. Kato Katz and Spontaneous Sedimentation techniques used in this study however allowed estimating a high overall prevalence of taeniosis reaching 2.4% and classifying these rural areas as hyperendemic [56] and at a higher rate when compared to the results reported previously in Madagascar [35,57]. The prevalence of taeniosis found in this study are quite similar to those previously described (0.1%-1.4%) in many African countries (Burundi, Congo, Ethiopia, Togo, Zambia, Cameroon, Kenya and Tanzania) and in some endemic countries of Latin America (Peru and Ecuador) using comparable coprological examination. Conversely these prevalence rates could be low compared to those reported in other African countries (Guinea, Nigeria, Ghana, and Gambia: up to 13%) [12,14,15,54] and those obtained using more sensitive diagnostic tools such as coproantigen ELISA and EITB methods (Zambia, Tanzania, Kenya and Democratic Republic of Congo: up to 23.4% [58][59][60][61]. No adult tapeworm carriers were detected in 4 out of the 12 villages investigated where participants presenting antibodies anti-Taenia solium and circulating antigens were however detected. Indeed, studies have shown that microscopy techniques are weakly sensitive missing 60-70% cases of taeniosis [62]. The best diagnostic assay for the intestinal taeniosis and constituting an effective tool for epidemiological studies would be the coproantigen detection ELISA having a sensitivity of about 95% and a specificity over 99% [62]. However these data have never been independently validated. This ELISA assay has been estimated to be two to ten times more sensitive than coprology methods [63]. But neither the corresponding test nor the necessary polyclonal antibodies for the coproantigen ELISA are available on the market and the method remains genus specific. Regarding molecular tools, a triplex Taq-Man probebased qPCR for the detection and discrimination of T.solium, T. saginata and T. asiatica in human stool [64] and the field application of a Loop-mediated isothermal AMPlification method (LAMP) for rapid identification of human taeniosis were also reported [65]. Using these tests especially the simple and sensitive LAMP technique in the context of Madagascar could be very useful in the field. Taeniosis cases reported in this study affected all age (mean age: 20.1 years) with 2.7% of prevalence in adults more than 15 years old. This population is not targeted by the national mass treatment programs (mainly treating children schooled between 5 to 15 years old). Thereby these results could guide the national program on mass chemoprevention (as primary intervention strategy against T. solium infections) to treat the entire population of all ages, especially the inhabitants of rural and remote areas.
Species of the Taenia spp. eggs identified in this study were all T. solium, the only species associated with cysticercosis. In order to analyze T. solium genotypes, we used cytochrome c oxidase subunit I which has been widely used for studying polymorphism and for establishing phylogenetic trees of Taenia genus (T. solium, T. saginata and T. asiatica) [41]. Our results confirm the findings of previous molecular analysis of T. solium worldwide, reporting that T. solium could be divided into two genotypes, Asian and African/American genotypes [66][67][68] (Fig 3). Our data strengthen the previous results of Michelet et al. and Yanagida et al. [40,41] showing that the two genotypes geographically disjunct are sympatric in Madagascar. Indeed confirmed by analyses of mtDNA and nuclear DNA, Malagasy people have mixed origins from Southeast Asia and from Southeast Africa [69]. Our results extend the previous findings that were made in Madagascar (on cysticerci from pigs) that the sympatric distribution of Asian and African-American T. solium genotypes is also found in the adult form and in the eggs of T. solium from human stool. The predominance of the Asian genotype (9/10 specimens) corroborated also to these prior results observed at a national level which occurred evenly at a District level of Ifanadiana for this study (Fig 3). Asian genotypes from this study are close to the Asian genotypes from India and Thailand but distant to the Asian genotypes from Indonesia, Korea and China as already reported [40,41], supporting the importance of the Indian influence on the diversity of people and culture in Madagascar [40, 69,70]. In this study, one of the two sequences obtained from the village of Sahavoemba resembles the Asian genotype but contains a base at the position 867 that is substituted (adenosine to guanine), as found in the African-American genotypes (Fig 3). Cross-fertilization and hybridization between individual worms possessing different genotypes has already been suggested by analyzing the nuclear genes [40]. Only one specimen from the village of Sahavoemba was an African-American genotype in our study. The sequence is closer to the other specimens isolated in Madagascar (Toliara) and Tanzania but distant to the sequences from Cameroon and Brazil (Fig 3).
The antigens used in this study based on lentil lectin affinity chromatography preparation [44] and Ag-ELISA (B158/B60) method [45] are used worldwide and constitute assays recognized by the WHO, both in sero-surveys for prevalence estimation and also for the diagnosis of individuals [50].
In this study the prevalence of antibodies (IgG) associated with an exposure with T. solium parasite detected by ELISA and by EITB methods using T. solium metacestode glycoproteins was quite similar (27.5% and 29.9% respectively). One study in Zambia has reported a higher seroprevalence (34.5%) than that found in this study using an EITB assay [71]. However this positivity rate is relatively high compared to those found in the general population or among villagers in other African countries using similar methods (ranging from 1.3% to 14.3%) and very high compared to the prevalence obtained in people with epilepsy (0 to 2.8%) [12][13][14][15]54]. The detection of cysticercosis-specific antibodies during field-based studies could help to identify areas of disease transmission [50] however they tend to over-estimate the prevalence of cysticercosis in sero-epidemiological surveys. Indeed, as antibodies disappear after 1 to 3 years in only 30-40% of seropositive people in endemic countries [72], some seropositivity will reflect past exposure to the parasite. In addition, infection in the general population could be in any tissue unlike epileptic individuals with NCC with cysts mainly localized in the Central Nervous System. Lower antibodies related to NCC may also be due to the lower sensitivity of EITB in parenchymal localization with few or calcified cysts [73] [74].
The presence of circulating antigens asserting an active cysticercosis case with viable cysts was observed at a prevalence of 12.4%. This prevalence is relatively high compared to those obtained in the following African countries (Cameroon, Zambia, Kenya, Burkina-Faso, Togo, and Senegal: 0.7% to 8.1%) [12][13][14]16,58] but extensively low compared to those found in Tanzania, Congo, and Togo (16.8% à 38%) [12,13]. However this positivity is similar to the prevalence of circulating Taenia antigens detected in Zambia, 12.5% [71].
Regarding the associated risk factors analysis, the use of soap for hand-washing was surprisingly found associated to active cysticercosis. A study using the Ag-ELISA test reported that participants who declared washing their hands by a dipping method and using the same water were more likely to be seropositive unlike those who used running water [59]. In our study villages, using soap for hand washing as a risk factor may be explained by the fact that people use the same contaminated river water for multiple purposes (washing vegetables, dishes, hands, etc). However other factors would be interesting to consider and could help understand this surprising result. The risk analysis result obtained with the variable "house water source" using standpipe and the probable protective factor when using the river with renewed running water seems to strengthen this previous report. In fact, clean water used directly as a source of water might be a protective factor of cysticercosis in this study.
Other risk factors tested significantly in this study are common factors classically reported in several previous investigations of T. solium infections in Africa and worldwide, such as open defection [54,71,[75][76][77][78], pork consumption (especially in preparation and cooking methods) [54,77,79] and increased age (adults >15 year old) [2,12,59,[80][81][82][83]. Many family members (�4) in the same household is a high risk insofar as a single T. solium carrier may easily cause cysticercosis to his family members through T. solium eggs shedding [11,84,85]. Among all identified factors, lack of knowledge of the population about soil-transmitted diseases and their risk (68.8% of the participants with elementary level of education) [16,75,86], the poor hygiene (52.1% of participants without latrines and practicing open defecation) and the isolation of these villages (� 6km from the main national road) promote the spread of T. solium [17,87]. Therefore, public health education in Madagascar should be a complementary approach for T. solium control. Indeed, these education and communication-based strategies have already been evaluated with success in several places like in rural communities of Burkina Faso [88], in an urban community of Mexico [54], through a computer-based educational health tool on T. solium in Tanzania and in Zambia [89,90], or also through a participatory community education workshop about T. solium in Peru [91]. Some protecting factors have nonetheless been identified as being a female, district versus local slaughter for pork source and habit of washing vegetables. Similar results were also obtained in village-community studies performed in Congo and in Burkina Faso that have reported a positive association between the presence of T. solium circulating antigens and being a male [2,61,92]. Likewise, several publications have reported that in the Districts where official meat inspection could take place, unlike local/home unregistered slaughtering (that has been largely shown to be an important source of transmission of T. solium infections), the prevalence of parasite carriage and associated infections decreased significantly [15,17,[93][94][95].
Our study may present limitations related to the number of samples analyzed. However, the epidemiological data obtained on the geographical distribution of T. solium taeniosis / cysticercosis in humans living in remote areas of Madagascar is important for identifying high-risk populations and applying control measures. This study identified and treated some tapeworm carriers of all age. Indeed, the treatment of taeniosis in humans is one of the basic "quick impact" interventions for strengthening the control/elimination of the parasite transmission and preventing the health burden associated with NCC. During this study, eight cases of active cysticercosis including three co-infections (taeniosis and active cysticercosis probably through self-infestation) were observed and clustered at household level with Taenia carrier. Some suggestive clinical manifestations of health burden associated with NCC (seizure in individuals aged between 20-49 years old [96]) were also recorded without being significantly associated with cysticercosis (Probably due to the high number of missing data: 76.3%) and without confirmation of diagnosis by neuroimaging. Indeed, neuroimaging is an Absolute-Major-Confirmative criteria of NCC. Nevertheless, important criteria of NCC according to the classification of Del Bruto and colleagues in 2017 (such as: specific anticysticercal antibodies/cysticercal antigens detected by immunodiagnostic tests, evidence of a household contact with T. solium infection, and clinical manifestations suggestive of NCC: mainly seizures) were recorded during this study. These data were reported to the local health authorities for the confirmation of the diagnosis of these cases and their eventual management.
Human health interventions in Madagascar remain challenging. Indeed, through a threeyear pilot project (2015-2017) in Madagascar, a significant reduction in the prevalence of taeniosis was observed after annual Mass Drug Administration (MDA) targeted adults and children >5 years old. However this reduction could not be maintained [97]. Investigations in pig health and environmental sector as part of a "one health" approach are therefore required to break the cycle of infection [98]. An ongoing project combining pig vaccination, drug treatment in pigs and in human tapeworm carrier will investigate this approach in Madagascar [99].

Conclusions
This is the first large-scale study in Madagascar to examine the prevalence of Taenia solium taeniosis/cysticercosis in humans and the association between potential risk factors measured at the individual-, household-, and village-level. Our results showed high rates of taeniosis cases affecting all ages in both townships investigated, strongly exposing human and pigs to cysticercosis by considering open defecation practices of the participants. Indeed, search of Taenia carrier among household contacts is currently recommended to identify the potential source of infection clustered within household and to reduce further spread/burden of the diseases. This high exposure of participants confirmed by the high rate of active cysticercosis cases should be managed with neuroimaging for a targeted and more efficient treatment. These data would be useful for guiding the authorities and all the entities involved in the programs and in the strategies for combating these diseases especially in rural and remote areas in Madagascar Island. Investigations in pigs and in environment would also be necessary as part of a "one health" approach to complete control strategies for those zoonotic neglected tropical diseases.

solium genotypes obtained in this study and in databases. Five T. solium
Asian genotypes (T. sol Asian I to V) and three for T. solium African-American genotype (T. sol Af/Am I to III) were aligned. T. solium Asian genotype cluster I sequence is shown as reference. Similarity is indicated with dot. Variant nucleotides are shown in lowercase and in capital letters if affecting amino acid sequence. Comparing to the Asian T. solium genotype cluster I (containing the majority of sequences obtained in this study), the Asian genotype cluster II (from Toamasina, Madagascar) showed one changed base (G instead of A) at the position 934 of the cox 1 completed sequence which changed the corresponding amino acid while the Asian genotype cluster III (from Indonesia) presented 3 substituted nucleotides (C, G and A changed into T, A and G: at the positions 650, 666 and 994 respectively). This last changed base was also the only variation observed in the Asian genotype cluster IV (from Nepal, Japan, China and Korea). The last Asian T. solium genotype obtained in this study, from Sahavoemba (Asian genotype cluster V) was also closely related to the majority of Asian T. solium genotype found in this study (Cluster I) except for one base: A modified in G at the position 867 which is a common substitution in the African-American T. solium genotype. The only African/ American genotype found in this study forming the African-American genotype cluster III (with Tanzania, Toliara Madagascar and Mexico) showed six substituted nucleotides (of which 3 modified amino acids) compared to the Asian genotype cluster II. The African-American genotype cluster II (Ecuador) and I (Mexico, Cameroon and Brazil) counted seven and nine substitutions respectively both with four changed amino acids. (TIF) S1 Table. Composition of the Townships and Fokontany (Ifanadiana District) investigated in this study. The geographic coordinates (GPS) of the 12 villages are indicated. (DOCX)