The authors have declared that no competing interests exist.
Conceived and designed the experiments: HHG AEG RHG. Performed the experiments: LMM VA GG AD SR. Analyzed the data: SEO HHG. Contributed reagents/materials/analysis tools: PPW VCWT. Wrote the paper: SEO HHG.
Neurocysticercosis is a leading cause of preventable epilepsy in the developing world. Sustainable community-based interventions are urgently needed to control transmission of the causative parasite, Taenia solium. We examined the geospatial relationship between live pigs with visible cysticercotic cysts on their tongues and humans with adult intestinal tapeworm infection (taeniasis) in a rural village in northern Peru. The objective was to determine whether tongue-positive pigs could indicate high-risk geographic foci for taeniasis to guide targeted screening efforts. This approach could offer significant benefit compared to mass intervention.
We recorded geographic coordinates of all village houses, collected stool samples from all consenting villagers, and collected blood and examined tongues of all village pigs. Stool samples were processed by enzyme-linked immunosorbent assay (ELISA) for presence of Taenia sp. coproantigens indicative of active taeniasis; serum was processed by enzyme-linked immunoelectrotransfer blot for antibodies against T. solium cysticercosis (EITB LLGP) and T. solium taeniasis (EITB rES33).
Of 548 pigs, 256 (46.7%) were positive for antibodies against cysticercosis on EITB LLGP. Of 402 fecal samples, 6 (1.5%) were positive for the presence of Taenia sp. coproantigens. The proportion of coproantigen-positive individuals differed significantly between residents living within 100-meters of a tongue-positive pig (4/79, 5.1%) and residents living >100 meters from a tongue-positive pig (2/323, 0.6%) (p = 0.02). The prevalence of taeniasis was >8 times higher among residents living within 100 meters of a tongue-positive pig compared to residents living outside this range (adjusted PR 8.1, 95% CI 1.4–47.0).
Tongue-positive pigs in endemic communities can indicate geospatial foci in which the risk for taeniasis is increased. Targeted screening or presumptive treatment for taeniasis within these high-risk foci may be an effective and practical control intervention for rural endemic areas.
Humans are the definitive host of the adult intestinal tapeworm, a condition known as taeniasis. People with taeniasis shed tapeworm eggs in their feces which contaminate the environment, particularly in rural regions where open defecation is common. When pigs are allowed to roam and consume human feces they are at risk of contamination with
Treatment of taeniasis is a key component of control and elimination strategies as adult intestinal tapeworms are the immediate common source of cysticercosis in both human and pigs. However, direct identification of taeniasis is complicated by low prevalence in endemic communities and by asymptomatic clinical course of infection
It is biologically plausible that pigs infected with a heavy-burden of
The hypothesis being tested is that the prevalence of taeniasis is higher among households in the immediate vicinity of a tongue-positive pig compared to households that are distant from the tongue-positive pig.
The study was conducted in the rural village of Rica Playa, in the Department of Tumbes, Peru. The northern coastal region of Peru is known to be endemic for
We conducted an initial census by visiting each household in the community and recording the age and sex of all resident household members, the number of pigs raised in the household, and the source of water and type of sanitary facilities available. We considered any individual who slept more than 2 nights per week in the village to be a resident. Latitude and longitude coordinates of each house were recorded using hand-held global positioning system (GPS) receivers (GeoExplorer II; Trimble, Sunnyvale, CA) with post-processed differential correction for sub-meter accuracy.
We then distributed a 500-ml plastic container with lid to all consenting residents ≥2 years old for collection of a whole stool sample. We also collected a 5-ml peripheral blood sample via venipuncture in standard serology vacuum tubes. Blood samples were maintained in coolers with ice-packs while in the field. All blood and stool samples were transported daily to the laboratory facility in Tumbes for processing.
Field teams captured all household pigs and a veterinarian inspected the tongue for nodules characteristic of
All samples were first processed in the laboratory facilities of the Global Health Center in Tumbes, Peru. Whole stool samples were examined macroscopically for the presence of
Fecal samples were concentrated by sedimentation then examined by light microscopy for the presence of
Human and pig sera samples were analyzed by enzyme-linked immunoelectrotransfer blot for presence of antibodies against
This study was reviewed and approved by the Institutional Review Board and the Institutional Ethics Committee for the Use of Animals at the Universidad Peruana Cayetano Heredia, Lima, Peru. All participants provided written informed consent, with parental or guardian consent required for the participation of minors. Treatment of animals adhered to the Council for International Organizations of Medical Sciences (CIOMS) International Guiding Principles for Biomedical Research Involving Animals.
We entered individual household coordinates into ArcMAP10 GIS software (ESRI; Redlands, CA) to generate a geo-referenced map of the community which included results of tongue-examination and laboratory exams. We used these coordinates to calculate the geodesic distance in meters from each household to the nearest house where a tongue-positive pig was raised.
All data were analyzed in STATA SE12 (StataCorp; College Station, TX). Fisher's exact test was used to compare distributions of proportions or to examine association between pairs of categoric measures. A score test was used to evaluate linear trend in log odds across categories. We constructed univariate logistic regression models with random-effects to estimate the odds of a positive test result while accounting for clustering at the household level. All tests were 2-sided, and significance was set at 0.05. We then used binomial family generalized estimating equations (GEE) with a log link and robust sandwich-type standard errors to estimate the population-averaged proportions of positive test results among household residents 1) living 100 meters or less from a house where a tongue-positive pig was found (includes index household) and 2) living more than 100 meters from a house where a tongue-positive pig was found. The 100-meter distance was chosen as a familiar measurement which could be readily applied in a community-based control intervention. We used households as the clustering variable and applied Quasilikelihood Information Criteria (QIC) in a STATA module to determine which variables to include
There were a total of 454 residents living in the study area at the time of the census including 240 (52.9%) males and 214 females (47.1%). The median age was 28 years (interquartile range [IQR] 14–47) with an overall range of 0–95 years. Residents were distributed among 101 different households (
Household characteristics, n = 101 | |
Average no. of houses per 100 meters radius | 7.8 |
No. residents per household, median [IQR] | 4 |
No. rooms per household, median [IQR] | 5 |
No. households with latrines, (%) | 46 (47.4) |
No. households raising pigs, (%) | 73 (80.2) |
No. pigs per household, median [IQR] | 4 |
No. households with ≥1 seropositive pig with more than 1 reactive band on EITB LLGP, (%) | 62 (61.4) |
No. households with ≥1 seropositive pig with 4–7 reactive bands on EITB LLGP, (%) | 8 (8.0) |
No. households with ≥1 tongue-positive pig, (%) | 7 (6.9) |
Distance in meters to nearest home with tongue-positive pig, median [IQR] | 448 [137–905] |
IQR = interquartile range.
Blood samples were obtained from 385 residents aged 2 years or older for a total blood sampling coverage of 84.8%. All 385 individuals who provided blood also provided a stool sample, and an additional 17 (3.7%) provided stool but no blood. A total of 402 residents aged 2 years or older provided a stool sample for 88.5% coverage of the total population. There were 52 (11.5%) unsampled individuals who provided neither blood nor stool, including 12 children <2 years old. People who did not provide samples were more likely to be males and also to raise pigs (data not shown). There were 47(10.4%) residents in 10 households who provided stool or blood samples but did not allow their pigs to be tested.
We captured 548 pigs in the village, of which 256 (46.7%) were positive for antibodies against one or more bands on EITB LLGP for cysticercosis (
EITB LLGP (1 or more reactive bands) | EITB LLGP (4–7 reactive bands) | Tongue exam | ||||
Variable | No. positive, (%) | p |
No. positive, (%) | p |
No. positive, (%) | p |
No. residents per household | ||||||
1–5 (1st tertile) | 147 (44.0) | 0.29 | 6 (1.8) | 0.89 | 5 (1.5) | 0.29 |
6–7 (2nd tertile) | 50 (51.0) | 1 (1.0) | 4 (4.1) | |||
8–10 (3rd tertile) | 59 (50.9) | 1 (0.9) | 2 (1.7) | |||
No. rooms per household | ||||||
1–4 (1st tertile) | 87 (41.8) | 0.07 | 4 (1.9) | 0.90 | 5 (2.4) | 0.92 |
5–6 (2nd tertile) | 112 (47.1) | 3 (1.3) | 4 (1.7) | |||
7–9 (3rd tertile) | 57 (55.9) | 1 (1.0) | 2 (2.0) | |||
No. houses per 100 meters | ||||||
1–5 (1st tertile) | 118 (47.4) | <0.01 | 3 (1.2) | 0.30 | 4 (1.6) | <0.01 |
6–9 (2nd tertile) | 85 (60.7) | 4 (2.9) | 7 (5.0) | |||
10–19 (3rd tertile) | 53 (33.3) | 1 (0.6) | 0 (0) | |||
Latrine in household | ||||||
Yes | 97 (46.4) | 0.93 | 5 (1.6) | 1.0 | 5 (2.4) | 0.76 |
No | 159 (46.9) | 3 (1.4) | 6 (1.9) | |||
No. pigs per household | ||||||
1–3 (1st tertile) | 30 (55.6) | 0.37 | 3 (5.6) | 0.04 | 0 (0) | 0.52 |
4–8 (2nd tertile) | 74 (46.8) | 2 (1.3) | 2 (1.3) | |||
9–26 (3rd terile) | 152 (45.2) | 3 (0.9) | 9 (2.7) |
Fisher's exact test, two-sided.
Of the 402 fecal samples, 6 (1.5%) were positive for the presence of
Green stars indicate the 5 households in which 6 coproantigen-positive individuals resided. A. Red circles indicate 100-meter ring radius around households where a tongue-positive pig was raised. There were 11 tongue-positive pigs raised in 7 different households (2 overlap on map). B. Blue circles indicate 100-meter ring radius around households where a seropositive pig with more than 4 reactive bands on EITB LLGP was raised. There were 8 pigs with 4+ reactive bands in 8 separate households.
EITB LLGP | EITB r33 | ELISA coproantigen | ||||
Variable | No. positive, (%) | p |
No. positive, (%) | p |
No. positive, (%) | p |
Sex | ||||||
Male | 75 (39.1) | 0.40 | 6 (3.1) | 0.16 | 2 (1.0) | 0.44 |
Female | 67 (34.7) | 13 (6.7) | 4 (2.0) | |||
No. residents per household | ||||||
1–5 (1st tertile) | 77 (41.4) | 0.06 | 7 (3.8) | 0.49 | 3 (1.6) | 0.29 |
6–7 (2nd tertile) | 38 (38.0) | 7 (7.0) | 3 (2.9) | |||
8–10 (3rd tertile) | 27 (27.3) | 5 (5.1) | 0 (0) | |||
No. rooms per household | ||||||
1–4 (1st tertile) | 52 (35.9) | 0.51 | 5 (3.5) | 0.45 | 1 (0.7) | 0.31 |
5–6 (2nd tertile) | 65 (35.5) | 10 (5.5) | 5 (2.6) | |||
7–9 (3rd tertile) | 25 (43.9) | 4 (7.0) | 0 (0) | |||
No. houses per 100 meters | ||||||
1–5 (1st tertile) | 46 (31.7) | 0.03 | 5 (3.5) | 0.46 | 2 (1.3) | 0.59 |
6–9 (2nd tertile) | 55 (47.0) | 8 (6.8) | 3 (2.5) | |||
10–19 (3rd tertile) | 41 (33.3) | 6 (4.9) | 1 (0.8) | |||
Latrine in household | ||||||
Yes | 69 (36.9) | 1.0 | 10 (5.4) | 0.82 | 2 (1.0) | 0.69 |
No | 73 (36.9) | 9 (4.6) | 4 (1.9) | |||
Pigs raised at household | ||||||
Yes | 126 (38.2) | 0.23 | 18 (5.5) | 0.50 | 5 (1.4) | 0.59 |
No | 16 (29.1) | 1 (1.8) | 1 (1.8) | |||
No. pigs per household | ||||||
1–3 (1st tertile) | 37 (31.6) | 0.06 | 8 (6.8) | 0.39 | 2 (1.6) | 0.87 |
4–8 (2nd tertile) | 48 (47.1) | 3 (2.9) | 2 (1.9) | |||
9–26 (3rd terile) | 41 (36.9) | 7 (6.3) | 1 (0.9) | |||
EITB LLGP positive pig in home | ||||||
Yes | 97 (39.4) | 0.69 | 16 (6.5) | 0.47 | 5 (1.9) | 1.0 |
No | 17 (36.2) | 1 (2.1) | 0 (0) | |||
Pigs not tested | 12 (32.4) | 1 (2.7) | 0 (0) | |||
Tongue positive pig in home | ||||||
Yes | 15 (57.7) | 0.09 | 2 (7.7) | 0.72 | 0 (0) | 1.0 |
No | 99 (37.1) | 15 (5.6) | 5 (1.79) | |||
Unknown | 12 (32.4) | 1 (2.7) | 0 (0) | |||
Distance to tongue positive pigs | ||||||
0–100 meters | 40 (52.0) | <0.01 | 8 (10.4) | 0.08 | 4 (5.0) | 0.02 |
101–500 meters | 30 (27.8) | 4 (3.7) | 0 (0) | |||
>500 meters | 72 (36.0) | 7 (3.5) | 2 (1.0) | |||
Tongue positive pig within 100 m radius | ||||||
Yes | 40 (52.0) | <0.01 | 8 (10.4) | 0.03 | 4 (5.0) | 0.02 |
No | 102 (33.1) | 11 (3.6) | 2 (0.62) |
Fisher's exact test, two-sided.
Of the 385 serum samples, 19 (4.9%) were positive for antibodies against
Of the 385 human sera samples, 142 (36.9%) were positive for antibodies against one or more bands on EITB LLGP for cysticercosis. The positive band distribution on EITB LLGP was as follows: 32 (22.5%) with a single band, 106 (74.6%) with 2–3 bands and 4 (2.8%) with 4–7 bands. In all 32 single-band positive samples the positive band corresponded to the 50 Kd glycoprotein (gp50). There was a significant trend for increasing seroprevalence over increasing categories of age (
Residents living >100 meters from household with tongue-positive pig |
Residents living ≤100 meters from household with tongue-positive pig n = 91 | ||||||
Laboratory assay | Prevalence (%) | 95% CI (%) | Prevalence (%) | 95% CI (%) | Crude PR | Adjusted PR |
95% CI |
EITB LLGP | 33.1 | 27.8–38.4 | 51.9 | 40.7–63.2 | 1.6 | 1.3 | 0.9–1.7 |
EITB rES33 | 3.6 | 1.5–5.7 | 10.4 | 3.5–17.3 | 2.9 | 3.1 | 1.2–8.3 |
ELISA coproantigen | 0.6 | 0–1.5 | 5.1 | 0.2–9.9 | 8.5 | 8.1 | 1.4–47.0 |
CI = Confidence interval.
PR = Prevalence ratio.
Generalized estimating equations (GEE) with household as the clustering variable and using robust sandwich-type standard errors. Best-fitting models for residents living within 100 meters of a tongue-positive pig included the following variables; 1) EITB LLGP; age and number of pigs raised within the house, 2) EITB r33; number of pigs raised within the house only, 3) ELISA coproantigen; age only.
Constructing 100-meter rings around pigs with 4+ bands instead of around tongue-positive pigs captures an additional coproantigen-positive individual (5/6; 83%). However, there was no statistically significant difference between the proportion of coproantigen-positive individuals living within 100-meters of a pig with 4+ bands (3/288, 1.0%) and residents living >100 meters from a pig with 4+ bands (3/114, 2.6%) (p = 0.4). Nor was the proportion of rES33-positive individuals living within 100-meters of a pig with 4+ bands (11/274, 4.0%) significantly different from the proportion of rES33-positive individuals living >100 meters from a pig with 4+ bands (8/111, 7.2%) (p = 0.2). The crude and adjusted prevalence ratios for these distance categories are shown in
Residents living >100 meters from household with tongue-positive pig |
Residents living ≤100 meters from household with tongue-positive pig n = 128 | ||||||
Laboratory assay | Prevalence (%) | 95% CI (%) | Prevalence (%) | 95% CI (%) | Crude PR | Adjusted PR |
95% CI |
EITB LLGP | 34.7 | 29.0–40.3 | 42.3 | 33.1–51.6 | 1.2 | 1.3 | 1.0–1.6 |
EITB rES33 | 4.0 | 1.7–6.4 | 7.2 | 2.4–12.1 | 1.8 | 1.9 | 0.7–5.3 |
ELISA coproantigen | 1.0 | 0–2.2 | 2.6 | 0–5.6 | 2.6 | 2.0 | 0.3–15.6 |
CI = Confidence interval.
PR = Prevalence ratio.
Generalized estimating equations (GEE) with household as the clustering variable and using robust sandwich-type standard errors. Best-fitting models for residents living within 100 meters of a seropositive pig with more than 4 reactive bands on EITB LLGP included the following variables; 1) EITB LLGP; age and number of pigs raised within the house, 2) EITB r33; age and number of pigs raised within the house, 3) ELISA coproantigen; number of pigs raised within the house only.
This study provides evidence of a strong geospatial association in endemic villages between pigs with a heavy-burden of
Other findings in our study support our conclusion. We observed a similar geospatial association between tongue-positive pigs and humans with serum antibodies against
Few other studies have examined geospatial patterns of
While we observed a strong geospatial association between taeniasis and tongue-positive pigs, the relationship was not absolute. There were two people with taeniasis in the community who were not associated with a nearby tongue-positive pig. There are several possible explanations. Recently acquired taeniasis may not be associated with visibly infected pigs, as the latent period between exposure to eggs and development of viable cysts is about 2 months
It is also important to note that we did not identify any cases of taeniasis occurring within the same household as a tongue-positive pig. This suggests that geographically-targeted screening for taeniasis should not be limited to the source household of the infected pig. Unrestrained pigs will roam beyond their immediate home to forage and may be exposed to
Although taeniasis/cysticercosis is considered a potentially eradicable disease, there has been relatively little data published on the effectiveness and sustainability of applied interventions. The TSOL 18 vaccine has been shown to be highly effective in protecting pigs from infection in community settings, but a commercial formulation is not yet available and there are lingering questions about villager uptake in a programmatic setting due to the requirement of multiple doses
As
Our study was conducted in a small rural village in northern Peru, a region in which
We thank the villagers and community leaders from Rica Playa for their participation and the staff of the Cysticercosis Elimination Program in Tumbes, Peru.