Antibody Banding Patterns on the Enzyme-Linked Immunoelectrotransfer Blot (EITB) Assay Clearly Discriminate Viable Cysticercosis in Naturally Infected Pigs

Enzyme-linked immunoelectrotransfer blot (EITB) detects antibodies against seven Taenia solium larvae antigens in three protein families (GP50, T24/42, and 8-kDa) with different structures and functions. EITB banding patterns against these antigens in pigs provide information about the course of infection and may discriminate viable cysticercosis. We analyzed the banding patterns and infection outcomes (presence of viable cysts, degenerated cysts, and any cysts) of 512 rural pigs. Banding patterns were grouped into homogenous classes using latent class analysis, and relationships with infection outcomes were assessed. Four classes were identified: 1 (n = 308, EITB-negative or positive for the GP50 family), 2 (n = 127, positive for GP50 (GP50 family), GP42-39 and GP24 (T24/42 family), but negative for 8-kDa antigens), 3 (n = 22, positive for GP50 and T24/42 antigens (GP42-39 and GP24), as well as to 8-kDa bands GP13, GP14, and GP18, but negative for GP21), and 4 (n = 55, positive for GP50 and T24/42 antigens, as well as to 8-kDa antigens GP21 and GP18 in combination). Pigs in classes 3 and 4 were more likely to have viable cysts (72.6% and 96.4%, respectively) than pigs in classes 1 and 2 (0.7% and 27.6%, respectively; p < 0.001). The number of infections with any cysts was higher in classes 3 and 4 (77.3% and 98.2%, respectively) and lower in classes 2 and 1 (34.7% and 4.9%, respectively; p < 0.001). Pigs with viable cysts represented >90% of pigs with any cysts in classes 3 and 4 (94.1% and 98.2%, respectively), while degenerated cysts were frequent among pigs with any cysts in classes 1, 3, and 2 (86.7%, 47.1%, and 43.2%, respectively; p < 0.001). EITB banding patterns strongly correlate with cysticercosis infection status in rural pigs, with classes 3 and 4 being more predictive of viable infections.


Introduction
Taenia solium taeniasis/cysticercosis is a zoonotic disease complex endemic in many low-and middle-income countries where informal pig husbandry and inadequate sanitation coexist [1,2].The lifecycle of T. solium includes humans as definitive hosts of the adult tapeworm (taeniasis) and pigs (as well as humans) as intermediate hosts of the tissue larval phase or cysticercus [3,4].Neurocysticercosis (NCC), the larval cyst infection of the central nervous system (CNS) is a major cause of epilepsy and other neurological morbidity in humans worldwide and accounts for approximately 30% of epilepsy cases in endemic areas [5][6][7].On the other hand, porcine cysticercosis negatively impacts the economy of small farmers due to the confiscation of infected pork, although in endemic areas, small farmers are often unaware of the infection of their pigs and communities do not participate in cysticercosis control [8][9][10].
The diagnosis of porcine cysticercosis is crucial for T. solium control strategies [11], as pigs with viable cysts are potential reservoirs of future tapeworms for humans.Given the relatively short lifespan of rural pigs, their infection rates often reflect the level of environmental contamination with T. solium eggs, and changes in disease prevalence in pigs may provide information in monitoring the success of control programs [12,13].Due to the strong geographic correlation between foci of cysticercotic pigs and the presence of human tapeworm carriers [14], small-radius clusters around cysticercotic pigs may also serve as an effective tool to identify individuals at high risk of taeniasis for timely diagnosis and treatment [15].However, a critical limitation is the lack of an affordable diagnostic test with adequate sensitivity and specificity to detect viable cysts in pigs in low-resource settings [16].Postmortem carcass dissection is the gold standard, but it is not feasible as a routine surveillance approach, in addition to being expensive and time-consuming in field studies [17].The tongue exam, although highly specific, is only moderately sensitive to severe infections and much less sensitive to mild infections, which tend to be substantially more frequent [18].Serological tests, targeting either antigen or antibody detection, have been widely used in epidemiological studies of porcine cysticercosis.However, major drawbacks include the poor ability of antibody detection tests to discriminate viable from non-viable infections [19], as well as the low specificity of antigen detection tests due to cross reactivity with Taenia hydatigena [20][21][22].
The enzyme-linked immunoelectrotransfer blot (EITB) assay detects antibody bands against seven T. solium glycoprotein antigens (GP50, GP42-39, GP24, GP21, GP18, GP14, and GP13) [23] and is the reference serological test for porcine cysticercosis due to its high initially reported levels of sensitivity and specificity (99% and 100%, respectively) [24].The presence of ≥1 EITB antibody bands has been classically interpreted as an indicator of exposure rather than viable infection [19,25].The high prevalence of EITB antibody bands in pigs from T. solium-endemic areas may arise from pigs with viable infections but may also represent transient responses in pigs exposed but not developing cysts, pigs with degenerated or past infections [26], pigs with maternal antibodies transferred by passive immunity [27], and pigs infected with T. hydatigena metacestodes (due to cross reactivity of the GP50 band) [28].The seven EITB antibody bands reflect responses against three distinct antigen families with defined structures and functions.The GP50 antigen family comprises transmembrane proteins that migrate at the 50 kDa position [29].The T24/42 antigen family comprises a set of proteins called tetraspanins (proteins involved in cell adhesion and proliferation processes) that migrate at 42-39 kDa and 24 kDa positions [30].The 8-kDa antigen family comprises small excretory/secretory proteins that migrate at 21, 18, 14, and 13 kDa positions and are eventually also found at the 24 kDa position on EITB [31].Categorizing positive or negative EITB results using optimal band-number cutoffs (≥3 bands) only moderately improves the ability to predict viable infections [32].In addition, EITB seroconversion in pigs can occur with diverse patterns of positive bands, which does not occur at random and can provide additional information about the likelihood of viable infection [33].
Unlike humans, the immune response profiles observed in the EITB banding patterns of pigs are influenced by several factors, such as their short lifespan, intensity of exposure, concomitant larval infections, and maternal immunity [34].In experimentally infected pigs, EITB results correlate with the stage of cyst viability (positive in pigs with viable cysts and negative in pigs without cysts or only with degenerated or caseous cysts) [35], whereas in naturally infected pigs, a proportional trend between the number of EITB bands and cyst burden was found [36].A previous study also found a correlation between EITB banding patterns and brain imaging findings in NCC patients, and some patterns were found to be highly predictive of viable infections [37].However, this association has not been properly assessed in naturally infected pigs yet.This paper reports the use of EITB banding patterns for discrimination of viable cysts in naturally infected pigs.

Study Design and Data Sources
This study is a cross-sectional assessment of the relationship between EITB banding patterns and viable cyst infection in naturally infected pigs involving the analysis of data (necropsy outcomes and EITB results) from three studies previously conducted in T. solium cysticercosis-endemic rural areas.The first study (year 2006) was a cysticercosis prevalence assessment in 326 pigs of ≥1 months of age from six rural communities in Tumbes, Peru [14].The second study (year 2012) involved a cohort of pigs that was part of a vaccination trial against porcine cysticercosis in Piura, northern Peru, including only unvaccinated control pigs ≥ 4 months (n = 104) who were raised under field conditions [38].The third study (years 2008-2012) included 83 adult pigs (≥12 months old) purchased from peasants in rural villages of Huancayo in the highlands of Peru (unpublished data, CWGP), considered a highly endemic area for human and porcine cysticercosis [39].Pigs from Huancayo were consecutively purchased among animals found to be positive upon tongue examination, whereas pigs from Tumbes and Piura were obtained from studies in community settings [14,38].
Pig necropsies were conducted using the procedures extensively applied in previous studies by our group [32,36].We defined the following infection outcomes separately: (1) presence of any cysts (viable or degenerated), (2) presence of viable cysts, and (3) presence of degenerated cysts.We used these criteria, since pigs with degenerated cysts (with or without viable cysts) may correspond with pigs with long-lived infections.We also obtained the proportion of pigs with viable cysts and degenerated cysts among pigs positive for any cysts.Parasite loads with viable and degenerated cysts were also recorded per pig and included as numerical variables.An EITB assay was performed as previously described by Tsang et al. [23] and adapted by Gonzalez et al. [24].Banding pattern readings were masked to the infection status of pigs.EITB results are reported as the presence or absence (1 and 0 values) of each of the seven EITB bands.These seven values, together, represent "EITB banding pattern" variable, indicating the combination of the positive bands.The pigs' sex and age (months) were also registered, and age was categorized as ≤8 and >8 months, as previously described [14,39].

Statistical Analysis
Pig characteristics (age, sex, infection outcomes, and EITB results) were described using summary statistics and compared by study site using bivariate analyses as required.The dichotomous results of each of the seven EITB bands were analyzed using latent class analysis (LCA) to group EITB banding patterns into homogenous groups ("latent classes") as previously described [37].We ran different models with increasing numbers of classes (2 to 6), using maximum likelihood methods to select the optimal number of classes according to a consensus between a statistical criterion (the model with the lowest Akaike and Bayesian information criteria (AIC and BIC) for model fit and parsimony, respectively) and the interpretability of classes (representative EITB banding patterns in each class).Pigs with unique EITB banding patterns (representing unusual responses) were excluded from LCA to avoid their inclusion as residuals in latent classes.Classes were described, and class prevalences ≥ 10% and conditional class membership probabilities for each band (close to 0 or 1) were also evaluated to confirm the model structure.Bivariate analyses between EITB classes and characteristics of the pig population were performed.Finally, we used logistic regression models to calculate odds ratios with 95% confidence intervals of viable cysticercosis in pigs according to EITB banding pattern classes and adjusted according to potential confounders (age categories, sex, and degenerated cyst infection).Bivariate models were applied for each covariate of interest.A second model was then applied, including all the covariates that were significant in bivariate analysis.Regression models used clustered-robust variances to account for the correlation of the data by study site.All the analyses were carried out in RStudio v1.2.5001 using the "PoLCA" package for latent class models.Statistical significance in the Wald test was set to 5%.

Ethics Statement
All the information described in this paper was obtained in previous studies by our group and reviewed and approved by The Institutional Ethics Committee for Animal Welfare Ethics of the School of Veterinary Medicine, Universidad Nacional Mayor de San Marcos (approval number 2004-007).All the procedures described in pigs were performed in accordance with the ethical principles outlined by the International Animal Care and Use Committee (IACUC).

Discussion
The diagnosis of porcine cysticercosis is challenging because of the difficulty and limitations of current diagnostic options that can be used in endemic settings [16,40].Necropsies are impractical in field studies, the tongue test is only moderately sensitive, and serological tests for antigen or antibody detection have low specificity due to cross reactions with other cestode infections [17,22] or cannot differentiate between infection and exposure [19].Our results demonstrate that the heterogeneity of EITB results can be grouped into four distinct EITB banding pattern classes, and among them, classes 3 and 4 are highly predictive of viable cyst infection (>75%), whereas less than 30% of pigs in class 2 had viable cyst infections, and an even lower proportion of pigs in class 1 had viable cysts.EITB banding patterns can substantially improve the discrimination of viable cysts in porcine populations.
A higher proportion of pigs with EITB banding pattern classes 3 and 4 had viable cysts and also had high cyst loads.These classes involved responses to GP50 and T24/42 antigens but also included responses to 8-kDa antigens that correlate with multiple viable parenchymal NCC and subarachnoid NCC in humans [37].The higher predictive value for viable swine cysticercosis of EITB classes 3 and 4 compared to class 2 responses can be explained by the presence of 8-kDa bands.These antigens are classified as excretory/secretory proteins [31,41], and therefore, the higher the number of viable cysts in pigs, the greater the production of circulating antigens and the more seroconversion, as reflected in class 3 and class 4 EITB responses.
Pigs with class 1 EITB banding patterns had an almost complete absence of viable infections, as only 2/308 pigs had viable cysts.Similarly, the presence of degenerated cysts in this class was lower (less than 5%) compared with classes 2, 3, and 4. A previous study showed that only 2/13 sera from human cases with a confirmed diagnosis of cysticercosis reacted to the GP50 band alone on EITB (false-positive rate of 84.6%) [42], whereas in a study of pigs experimentally infected with T. hydatigena, the presence of the GP50 band alone indicated cross reactivity against T. hydatigena metacestode antigens [28], highlighting the limited specificity of the GP50 band alone in rural pigs (which may be exposed to a higher number of parasites).
Viable cysts were observed in less than 30% of pigs with class 2 responses, and degenerated cyst infections represented 43.2% of pigs with any cysts in this class, reflecting the persistence of antibody responses to antigens GP42-39 and GP24, which are frequent in class 2 (T24/42 antigen family) [23,43], which may explain the presence of antibodies to these antigens in pigs with degenerated or resolved infections [34].However, some class 2 pigs with three bands (GP50, GP42-39, and GP24) had a high cyst burden.It is possible that these pigs may also express 8-kDa antigens, which can also migrate to the GP24 band on EITB [31].In any case, responses of class 2 are not highly predictive of viable infection and should also be considered with caution for diagnosis of viable cysticercosis in naturally infected pigs.
We also observed that pigs older than 8 months had greater odds of having viable cysts in the regression model, and EITB class 3 and 4 responses were also more frequent in pigs older than 8 months.These results are consistent with previous studies [36] and suggest that older pigs have been more exposed to T. solium eggs through their lifetime, with more time to develop cysts and a stronger antibody response than younger pigs [39,44,45].
In our study, LCA identified four EITB banding patterns that reflected characteristic responses against T. solium antigen families.Class 1 includes negative pigs and responses to GP50 antigens alone, while class 2 includes responses to T24/42 antigens, and classes 3 and 4 include responses to GP50, T24/42, and 8-kDa antigens.EITB classes 1 and 2 were more frequent in the total pig population, since GP50, GP42-39, and GP24 antigens belonging to these classes are immunodominant and apparently induce a more persistent immune response [23,46].Classes 3 and 4 are less common in the pig population [39].Classes 3 and 4 showed a similar structure, with both classes including responses to 8-kDa antigens.However, LCA recognizes them as distinct, separate classes that present different clinical and a somewhat different sex/age profiles.Pigs with class 4 responses were also more frequent in Huancayo, where pigs were selected using a tongue test, while EITB classes 1 and 2 were more frequent in pigs from community studies in Tumbes and Piura.
Our findings should be interpreted in light of certain potential drawbacks.We did not determine potential confounders such as the presence of other cestode infections (e.g., T. hydatigena), the inclusion of which in the regression model could have provided a more precise estimate of the association between EITB banding patterns and infection outcomes.The use of data from three studies with differences regarding EITB results and infection outcomes may affect the generalizability of our findings.For this reason, we used clustered robust variances that account the correlation of the data by study site in the regression model.Finally, the use of latent classes as predictors in the logistic regression model results in a certain degree of uncertainty associated with a measurement error in the predictor variable, which can produce a bias in the estimation of the regression parameters.Finally, band pattern readings are qualitative rather than quantitative and depend on the expertise of readers, so further studies using quantitative methods, such as Luminex, should more precisely assess the correlation between EITB responses and infection status in porcine cysticercosis.

Conclusions
The accurate diagnosis of porcine cysticercosis, the intermediate reservoir of T. solium infections, is a key component for the effectiveness of control strategies.EITB banding patterns correlate strongly with cyst infection status in pigs, and their interpretation improves the diagnosis of cysticercosis in porcine populations.EITB class 3 and class 4 responses are highly predictive of viable cyst infections, whereas class 2 responses are associated with a combination of viable and non-viable cyst infections, and class 1 responses usually represent pigs without cysts.Even if EITB serves as the reference serological test for porcine cysticercosis, its implementation in rural settings is difficult and not affordable.New, easily applicable assays are required, and the availability of recombinant antigens produced from native antigens in rapid assays (for example, using antigens with greater predictive value and their respective banding patterns) may improve the real-time diagnosis of porcine cysticercosis in T. solium-endemic areas.

Supplementary Materials:
The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/pathogens13010015/s1,Table S1: Goodness-of-fit statistics to select the optimal number of classes for EITB banding patterns in latent class analysis; Table S2 Informed Consent Statement: Not applicable.

Data Availability Statement:
The datasets collected for this study are available from the corresponding author upon request.
: Conditional class-membership probabilities for each EITB band and class prevalences obtained in the 4-class model.Author Contributions: Conceptualization, G.A. and H.H.G.; methodology, G.A. and A.G.L.; data curation, G.A., C.M.J. and A.E.G.; formal analysis: G.A. and A.G.L.; supervision: H.H.G., A.G.L. and S.E.O.; writing-original draft preparation, G.A.; writing-review and editing, A.G.L., H.H.G., S.E.O., C.M.G., T.L.-U., M.A.-G., L.A.G.-P., J.A.B. and C.M.J.; funding acquisition, A.G.L., A.E.G. and H.H.G.All authors have read and agreed to the published version of the manuscript.Funding: Datasets used in this paper were previously obtained from studies supported by the Bill & Melinda Gates Foundation through grants 23981, 438, and 1016506.G.A. was partially supported by the Fogarty International Center/NIH.(training grants D42TW001140 and D43TW007393).A.G.L. is also supported by NIH training grant D43TW007393.Institutional Review Board Statement: This study was reviewed and approved by the Institutional Ethics Committee for Animal Welfare Ethics of San Marcos University (approval number: 2004-007) and conducted in accordance with the ethical principles outlined by the International Animal Care and Use Committee (IACUC).

Table 1 .
Characteristics of the total pig population and distribution by study site.

Table 2 .
EITB class description and EITB banding patterns in each class.

Table 3 .
Bivariate comparisons between pig characteristics and EITB banding pattern classes *.

Table 4 .
Odds ratios of having viable cysticercosis according to pig characteristics obtained from logistic regression models *.