Evaluation of a multiplex assay and two commercial ELISAs for the serological detection of Actinobacillus pleuropneumoniae using Bayesian latent class modelling

Evaluation of a multiplex assay and two commercial ELISAs for the serological detection of Actinobacillus pleuropneumoniae using Bayesian latent class modelling | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Evaluation of a multiplex assay and two commercial ELISAs for the serological detection of Actinobacillus pleuropneumoniae using Bayesian latent class modelling Katharine R. Dean, Kari Lybeck, Ingunn Anita Samdal, Anniken Jerre Borge, and 6 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-9608302/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Actinobacillus pleuropneumoniae (APP) remains a major cause of respiratory disease in swine and accurate serological tools are essential for surveillance, particularly in sub-clinical herds and low prevalence settings. We evaluated the diagnostic performance of a microsphere-based multiplex fluorescent immunoassay (MFIA) and two commercial ELISAs (1–12 ELISA and ApxIV ELISA) using a Bayesian latent class model and field samples from Norwegian swine herds. Although both ELISAs and the MFIA showed high specificity, the ApxIV ELISA demonstrated lower sensitivity (89.4%) than the other assays. These findings align with previously reported variability in the sensitivity of the ApxIV-based ELISA. The MFIA showed high sensitivity, demonstrating the potential of multiplex immunoassays for APP serology. We also explored optimization of the ApxIV ELISA cut-off, which indicated that lowering the threshold could improve the sensitivity of the test without compromising the specificity for our samples. Our estimates can be influenced by several factors, including the serotype diversity in the population and assumptions related to latent class modelling. Despite these constraints, our findings support the use of MFIA and the ELISAs as accurate tools for APP surveillance, while highlighting that the ApxIV ELISA may be less suitable as a stand-alone screening test. These results underscore the importance of validating diagnostic assays in populations where they will be applied to ensure that they are fit-for-purpose. APP BLCM test evaluation diagnostic accuracy sensitivity specificity Figures Figure 1 Figure 2 Figure 3 Figure 4 Highlights • MFIA and 1–12 ELISA showed high sensitivity and specificity for APP surveillance • ApxIV ELISA had lower sensitivity and may be less suitable for screening • Lowering the ApxIV cut-off could improve sensitivity without loss of specificity • APP surveillance requires fit-for-purpose test selection based on objectives 1. Introduction Actinobacillus pleuropneumoniae (APP) is a contagious respiratory pathogen in swine and the causative agent of porcine pleuropneumonia (Gottschalk, 2012 ). The disease is associated with significant economic losses for pig producers from increased mortality, antimicrobial and vaccination costs, and decreased growth performance (Stygar et al., 2016 ). APP is a genetically diverse bacterial species with 19 known serotypes recognized worldwide (Arnal Bernal et al., 2024 ; Kwan et al., 2025 ; Paulina and Dawid, 2025 ; Stringer et al., 2021 ). Serotype specificity is determined primarily by capsular polysaccharides (CPS) and, in some serotypes, the O chain portion of lipopolysaccharide (LPS) (Blackall et al., 2002 ; Boekema et al., 2004 ; Dubreuil et al., 2000 ). Cross-reactive antigenic determinants in the LPS O chain form the basis for serological grouping, notably serotypes 1/9/11, 3/6/8/15, and 4/7 (Gottschalk, 2012 ). APP virulence is closely linked to its set of repeat-in-toxins (RTX) toxins (Boekema et al., 2004 ; Bossé et al., 2002 ). Different serotypes produce distinct combinations of Apx toxins: ApxI, ApxII, ApxIII, or mixtures such as ApxI/II and ApxII/III. In contrast, ApxIV is produced exclusively by APP and by all known serotypes, making it a highly specific antigenic marker for the species (Dreyfus et al., 2004 ; Schaller et al., 2001 ). Clinically, APP infection results in severe fibrinous and necrotizing pleuropneumonia, causing high morbidity and mortality in acutely affected herds (Sassu et al., 2018 ). Pigs that survive the acute phase—or that are infected sub-clinically—can become long‑term asymptomatic carriers, acting as reservoirs for within‑herd transmission (Sassu et al., 2018 ). The presence of such carriers complicates eradication efforts and contributes to the persistence of APP despite prevention and control measures. Serological surveillance plays a key role in detecting and monitoring APP in sub-clinically infected herds (Gottschalk, 2015 ; Sassu et al., 2018 ). Several assays have been developed to identify antibodies against APP antigens, including the complement fixation (CF) test, enzyme-linked immunosorbent assays (ELISAs), and microsphere-based multiplex immunoassays (Berger et al., 2017 ; Gottschalk, 2015 ). The CF test is no longer recommended due to poor sensitivity and specificity for APP detection (Gottschalk, 2015 ). Instead, commercial and in-house ELISAs have become primary diagnostic serological tools. ApxIV-based ELISAs exploit the species-specific and universally expressed ApxIV toxin. However, reduced sensitivity has been reported, potentially due to ISApl1-mediated inhibition of ApxIV expression in certain strains (Eamens et al., 2012 ; Opriessnig et al., 2013 ; Tegetmeyer et al., 2008 ). LPS-based ELISAs detect antibodies targeting serotype-specific LPS O-chain structures (Gottschalk, 2012 ; Gottschalk et al., 1997 ). These assays can detect antibodies to individual serotypes or serogroups, and are considered highly sensitive and specific, but they can be costly when multiple serotypes must be tested (Broes et al., 2007 ). Additionally, some ELISAs have been based on CPS or common surface proteins (Gottschalk 2012 ). More recently, microsphere‑based multiplexed fluorometric immunoassays (MFIA) have been developed, allowing simultaneous detection and differentiation of antibodies against multiple APP serotypes within a single well (Berger et al., 2017 ; Caya et al., 2014 ). These assays use the Luminex Multi-Analyte Profiling of analyte x (xMAP) technology, where fluorescently barcoded microspheres are conjugated to serotype‑specific antigens. Fluorescence is quantified using a flow cytometer, enabling high‑throughput testing. Each bead’s identity is resolved by its unique internal dye ratio, while antibody binding is detected through phycoerythrin (PE) fluorescence, measured by a secondary laser, allowing simultaneous quantification of multiple analytes (Berger et al., 2017 ; Christopher-Hennings et al., 2013 ). Although promising, the multiplex assays for APP have not yet undergone extensive validation using field samples. Diagnostic test accuracy describes how well a test discriminates between infected and non-infected individuals, commonly measured through sensitivity, specificity, and predictive values. For APP, no perfect reference test exists for detecting antibodies in sub-clinically infected pigs (Sassu et al., 2018 ). Consequently, earlier studies have relied mainly on experimentally infected animals or animals of known status, expert classification, or comparison between multiple imperfect tests (Berger et al., 2017 ; Caya et al., 2014 ; Costa et al., 2011 ; Dreyfus et al., 2004 ; Eamens et al., 2012 ; Gottschalk et al., 1994 ; Klausen et al., 2007 ; Opriessnig et al., 2013 ). Bayesian latent class analysis has been underutilized for estimating the performance of APP diagnostic methods, despite being well-suited for evaluating test performance in the absence of a perfect reference test (Enøe et al., 2001 ). At the Norwegian Veterinary Institute, the ID Screen APP Screening Indirect (serotypes 1–12) test from Innovative Diagnostics (Grables, France) is used for routine screening for APP, followed by retesting of positive samples using the IDEXX APP-ApxIV Ab Test from IDEXX Laboratories (Westbrook, Maine, USA). For the export of live animals, industry partners have been advised to screen for APP using IDEXX APP-ApxIV Ab only. Because the ID Screen 1–12 ELISA and the ApxIV ELISA detect antibodies against different antigens, animals in infected herds may test positive in only one of the two assays (Gottschalk, 2015 ). Disagreement between the tests can make interpreting the results challenging. Building on the successful population-wide eradication of Mycoplasma hyopneumoniae (Gulliksen et al., 2021 ), the Norwegian pig production sector has placed growing emphasis on improving herd health by encouraging piglet producers and finishers to convert to Specific Pathogen Free (SPF) herds (Norsvin SA, 2026 ). These SPF herds are managed under stricter health and biosecurity standards to ensure that they remain free from several major pathogens, including APP. Conversion of a conventional pig herd to SPF status typically involves planned depopulation, thorough cleaning and disinfection, and repopulation with SPF animals after a vacancy period of about three weeks. Certification depends on a structured testing program, and once certified, continuous health monitoring is required, with different protocols for breeding and commercial herds. There are currently no international standards for how APP freedom should be established and maintained, which diagnostic tests should be used, and under what conditions (Gottschalk, 2015 ; Sassu et al., 2018 ). Consequently, improving knowledge about the diagnostic accuracy of available screening tests is essential, as it directly determines the reliability and economic costs associated with declaring herds APP-free under different sampling strategies. To address this research gap, the aim of this study was to estimate the sensitivity, specificity, and predictive values of a microsphere-based multiplex fluorescent immunoassay and two commercial ELISAs for the detection of antibodies against APP in Norwegian swine herds. 2. Material and methods 2.1 Study population and sampling The study was primarily conducted in Rogaland, a county in southwest Norway, shown in Fig. 1 . Rogaland lies in the region of Norway with the highest number and density of pig herds (Grøntvedt et al., 2023 ). Norwegian pig production mainly consists of independent producers that are organized into a hierarchical breeding pyramid, with specialized herds for different production stages. APP is presumed to be prevalent in herds throughout the production pyramid except for herds with approved Specific Pathogen Free (SPF) status, and a previous study has highlighted the importance of APP in porcine acute respiratory disease in Norway (Cohen et al., 2020 ). Serum samples from 20 sow herds and one genetic nucleus herd were originally collected for other reasons and later re-used for this study. Samples from the sow herds were collected in 2022 as part of the ‘Små-i-Ro' project to investigate the association between APP and abattoir recordings (Midtveit, 2023 , 2022 ). The 20 herds from ‘Små-i-Ro' represented the 10 sow herds with the highest and lowest abattoir recordings for pleuritis and pericarditis through routine meat inspection from five slaughterhouses in Rogaland. Samples from a genetic nucleus herd in Vestfold were collected in 2024 as part of routine screening of breeding animals that were 8–10 weeks old. We assumed that the genetic nucleus herd, the herds with high remarks, and the herds with low remarks had different prevalence for APP and considered them three separate populations for the analysis. Blood samples from the selected herds were collected from v. jugularis into serum tubes (without anticoagulant) and transported to the laboratory for centrifugation (2500 x g for 5 min) and further processing. Serum was harvested and frozen at − 20°C, until analysis. 2.2 Diagnostic tests 2.2.1 Microsphere-based multiplex fluorescent immunoassay (MFIA) A multiplex method with seven sets of APP antigens coupled to magnetic beads, named Swinecheck MP APP 1-9-11, 2, 3-6-8-15, 4–7, 5, 10 & 12 from Biovet (Saint-Hyacinthe, QC, Canada) was tested on the swine serum samples. The magnetic beads are coupled to purified long-chain-LPS antigens from APP serotypes/groups for the detection of IgG antibodies against APP1-9-11, APP 2, APP 3-6-8-15, APP 4–7, APP 5, APP 10 and APP 12. The multiplex assay was run according to the manufacturer’s instructions with minor modifications. Serum samples were diluted 1/100 in the sample buffer and mixed well. A mix of seven magnetic beads coupled with APP antigens (Table 1 ) were vortexed (5s), sonicating (5s) and 50 µL was dispensed in each well on a 96 well plate (Bio-Plex Pro Flat Bottom Plates from BioRad (Hercules, CA, USA)). Then, 50 µL of samples, positive and negative controls were added to dedicated wells and incubated 60 min at 800 rpm in dark on a MixMate shaker (Eppendorf, Hamburg, Germany). The beads on the plate were washed 3x with the supplied wash buffer using Bio-Plex ProWash Station with magnetic separator (Bio-Rad, Hercules, CA, USA) between steps. Primary antibody (100 µL) was added to each well, incubated for 30 min on shaker and then washed. Secondary antibody with R-phycoerythrin (PE) (100µL) was added to the wells, incubated for 30 min on the shaker and then washed again. The plate was analysed by applying 125 µL wash buffer to the wells, resuspended on shaker for 2 min, and read in Bio-Plex 200 (Bio-Rad, Hercules, CA, USA) using Bio-Plex Manager Software version 6.2. The signal was measured as median fluorescence intensity (MFI) of at least 50 beads per bead region, and the doublet discriminator gate was set to 5,000–25,000. The blank signal was subtracted from the sample signal. The controls were used to calculate the sample-to-positive ratio (S/P) using the following formulae: $$\:\frac{\text{S}}{\text{P}}\:=\frac{\text{M}\text{F}\text{I}\text{s}\text{a}\text{m}\text{p}\text{l}\text{e}\:-\:\text{M}\text{F}\text{I}\text{n}\text{e}\text{g}\text{a}\text{t}\text{i}\text{v}\text{e}\:\text{c}\text{o}\text{n}\text{t}\text{r}\text{o}\text{l}}{\text{M}\text{F}\text{I}\text{p}\text{o}\text{s}\text{i}\text{t}\text{i}\text{v}\text{e}\:\text{c}\text{o}\text{n}\text{t}\text{r}\text{o}\text{l}\:-\:\text{M}\text{F}\text{I}\text{n}\text{e}\text{g}\text{a}\text{t}\text{i}\text{v}\text{e}\:\text{c}\text{o}\text{n}\text{t}\text{r}\text{o}\text{l}}\:$$ For the analysis, inconclusive results were considered positive. Table 1 Overview over antigens coupled to beads and their cut offs Microsphere Mix Constituent Bead Code (Region) S/P cut-offs Negative Inconclusive Positive APP 1-9-11 antigen 51 < 0.40 0.40 ≤ S/P < 0.55 ≥ 0.55 APP 2 antigen 52 < 0.30 0.30 ≤ S/P < 0.40 ≥ 0.40 APP 3-6-8-15 antigen 53 < 0.30 0.30 ≤ S/P < 0.40 ≥ 0.40 APP 4–7 antigen 54 < 0.40 0.40 ≤ S/P < 0.50 ≥ 0.50 APP 5 antigen 55 < 0.30 0.30 ≤ S/P < 0.40 ≥ 0.40 APP 10 antigen 56 < 0.30 0.30 ≤ S/P < 0.40 ≥ 0.40 APP 12 antigen 57 < 0.20 0.20 ≤ S/P < 0.30 ≥ 0.30 2.2.2 ELISAs Two commercial ELISA kits were used to test the swine sera for comparison. The kits were the ID Screen APP Screening Indirect (serotypes 1 through 12) from Innovative Diagnostics (Grables, France), hereafter called the ‘1–12 ELISA’, and the IDEXX APP-ApxIV Ab Test from IDEXX Laboratories (Westbrook, ME, USA), hereafter called the ‘ApxIV ELISA’. The 1–12 ELISA is designed for the detection of IgG antibodies against APP serotypes 1 through 12 in swine serum, plasma, or meat juice, and uses partially extracted, native polysaccharidic surface antigens derived from the different serotypes (Innovative Diagnostics, 2026 ). The ApxIV ELISA, which employs a recombinant ApxIV antigen, is intended for the detection of IgG antibodies against APP in swine serum and plasma. Both kits were run according to the manufacturer’s instructions. In both ELISAs samples were tested at 1/10 dilution and absorbances were read at 450 nm. The controls were used to calculate the sample-to-positive ratio (S/P%) using the following formula: $$\:\frac{\text{S}}{\text{P}}\text{%}\:=\left(\frac{\text{O}\text{D}\text{s}\text{a}\text{m}\text{p}\text{l}\text{e}\:-\:\text{O}\text{D}\text{n}\text{e}\text{g}\text{a}\text{t}\text{i}\text{v}\text{e}\:\text{c}\text{o}\text{n}\text{t}\text{r}\text{o}\text{l}}{\text{O}\text{D}\text{p}\text{o}\text{s}\text{i}\text{t}\text{i}\text{v}\text{e}\:\text{c}\text{o}\text{n}\text{t}\text{r}\text{o}\text{l}\:-\:\text{O}\text{D}\text{n}\text{e}\text{g}\text{a}\text{t}\text{i}\text{v}\text{e}\:\text{c}\text{o}\text{n}\text{t}\text{r}\text{o}\text{l}}\right)\:\times\:100$$ Recommended S/P% cut offs for the two ELISAs are shown in Table 2 . Inconclusive results were considered positive for the evaluation. Table 2 APP ELISAs and their cut-offs ELISA kit S/P% cut-offs Negative Inconclusive Positive 1–12 ELISA ≤ 25 25 ≤ S/P% < 30 ≥ 30 ApxIV ELISA < 40 40 ≤ S/P% < 50 ≥ 50 2.3 Bayesian latent class analysis We evaluated the performance of the tests using a Bayesian latent class model for three tests and three populations. The analysis uses a latent class to represent an individual’s true status for APP, which is not directly observed in the data without a perfect reference test. We defined the latent condition as infection with or previous exposure to APP serotypes 1–12 and 15. We grouped the herds to form three populations consisting of those with the highest remarks, the lowest remarks, and the genetic nucleus herd, based on the assumption that they would have differing APP prevalence. We specified a Bayesian latent class model based on the Hui-Walter framework (Hui and Walter, 1980 ). We assumed that the test characteristics were constant across populations and that the tests were conditionally independent given the infection status. For each population, \(\:k\) , we summarized the test results as a vector, \(\:{y}_{k}\) , with the number of individuals in each possible combination of test outcomes. We used a multinomial distribution to model the observed data as follows: $$\:\left.{y}_{k}\right|{\pi\:}_{k}{Se}_{1},{Se}_{2},\:{Se}_{3,}{Sp}_{1},{Sp}_{2},{Sp}_{3}∼Multinomial\left({p}_{k},\:{n}_{k}\right),\:k=\text{1,2},3$$ $$\:{p}_{000k}=\:{\pi\:}_{k}(1-{Se}_{1})(1-{Se}_{2})(1-{Se}_{3})+(1-{\pi\:}_{k}){Sp}_{1}{Sp}_{2}{Sp}_{3}$$ $$\:{p}_{100k}=\:{\pi\:}_{k}{Se}_{1}(1-{Se}_{2})(1-{Se}_{3})+(1-{\pi\:}_{k})(1-{Sp}_{1}){Sp}_{2}{Sp}_{3}$$ $$\:{p}_{010k}=\:{\pi\:}_{k}(1-{Se}_{1}){Se}_{2}(1-{Se}_{3})+(1-{\pi\:}_{k}){Sp}_{1}(1-{Sp}_{2}){Sp}_{3}$$ $$\:{p}_{110k}=\:{\pi\:}_{k}{Se}_{1}{Se}_{2}(1-{Se}_{3})+(1-{\pi\:}_{k})(1-{Sp}_{1})(1-{Sp}_{2}){Sp}_{3}$$ $$\:{p}_{001k}=\:{\pi\:}_{k}(1-{Se}_{1})(1-{Se}_{2}){Se}_{3}+(1-{\pi\:}_{k}){Sp}_{1}{Sp}_{2}(1-{Sp}_{3})$$ $$\:{p}_{101k}=\:{\pi\:}_{k}{Se}_{1}(1-{Se}_{2}){Se}_{3}+(1-{\pi\:}_{k})(1-{Sp}_{1}){Sp}_{2}{(1-Sp}_{3})$$ $$\:{p}_{011k}=\:{\pi\:}_{k}(1-{Se}_{1}){Se}_{2}{Se}_{3}+(1-{\pi\:}_{k}){Sp}_{1}(1-{Sp}_{2}){(1-Sp}_{3})$$ $$\:{p}_{111k}=\:{\pi\:}_{k}{Se}_{1}{Se}_{2}{Se}_{3}+(1-{\pi\:}_{k})(1-{Sp}_{1}){(1-Sp}_{2}\left)\right(1-{Sp}_{3})$$ where, \(\:{\pi\:}_{k}\) , is the prevalence of APP in each population, and \(\:Se\) and \(\:Sp\) are the sensitivity and specificity of each test. We used Bayesian analysis to estimate the unknown parameters, which were the sensitivity and specificity of each test, as well as the true prevalence in the populations. Since we lacked knowledge about the true values of these parameters, we assigned minimally informative \(\:Beta\left(\text{1,1}\right)\) distributions as priors. We estimated the posterior distributions by running two independent Monte-Carlo Markov chains (MCMC) with dispersed starting values for a total of 200,500 iterations, including a 5,000 burn-in and 1/20 thinning, for a total of 20,000 posterior samples per chain. The effective sample size was above 1000 for all variables. We evaluated model convergence by visually inspecting the MCMC trace plots and checking that the Gelman–Rubin potential scale reduction factor was below 1.05 for all parameters. We calculated the mean and median values for the estimated parameters and the 95% credible intervals from the posterior samples (PCI). We used the samples from the MCMC chains to calculate the sensitivity and specificity of the serial and parallel interpretations of the different paired test combinations and the positive and negative predictive values. We implemented the model in JAGS version 4.3.1 (Plummer, 2025 ) and R version 4.4.3 (R Core Team, 2025 ) using the rjags (Plummer, 2025 ), runjags (Denwood, 2016 ), dplyr (Wickham et al., 2023 ), tidyr (Wickham et al., 2019 ), and ggplot2 (Wickham, 2016 ). The software code and data for reproducing the results is available on Github: https://github.com/NorwegianVeterinaryInstitute/App_test_evaluation . The manuscript was prepared using the “STARD-BLCM: Standards for the reporting of diagnostic accuracy studies that use Bayesian Latent class models” checklist (Kostoulas et al., 2017 ). 2.4 Sensitivity analysis As all three tests measured IgG antibodies towards APP, we investigated the assumption of independence between tests as a sensitivity analysis. We compared an alternative model with conditional dependence between tests to the original model without conditional dependence. We used minimally informative \(\:Beta\left(\text{2,1}\right)\) distributions as priors for the sensitivity and specificity, and \(\:Beta\left(\text{1,1}\right)\) for the prevalence. 2.5 Optimizing the cut-off values for ApxIV ELISA Previous studies have found lower sensitivity for the ApxIV ELISA, suggesting that the manufacturer’s recommended cut-off is too high (Eamens et al., 2012 ; González et al., 2017 ; Opriessnig et al., 2013 ). Therefore, we estimated the optimal cut-off value for the ApxIV ELISA for our samples using the posterior probability as described in (Olsen et al., 2022 ). Following Olsen (2022), we defined the optimal cut-off value as the maximum sum of the sensitivity and specificity. To explore the performance of the test across a range of cut-off values, we constructed a receiver operator curve (ROC) with 95% PCI. We reran the three test-three population model without covariance using the new cut-off value for the ApxIV ELISA test. 3. Results 3.1 Descriptive statistics We tested a total of 284 samples with the three tests: 1–12 ELISA, MFIA, ApxIV ELISA. When using the recommended cut-off values, 188 (66.2%) samples were positive in the 1–12 ELISA, 188 (66.2%) were positive in the MFIA, and 171 (62.3%) were positive in the ApxIV ELISA. For 256 (90.1%) samples, there was agreement between all three tests, whereas for 28 (9.9%) samples, the results differed. Table 3 shows the paired test outcomes based on the recommended cut-off values for each test and subpopulation. Table 3 Paired test outcomes for each subpopulation with the 1–12 ELISA, MFIA, and ApxIV ELISA, using the manufacturers recommended cut-off values. Population Test results for 1–12 ELISA / MFIA / ApxIV ELISA -/-/- +/-/- -/+/- +/+/- -/-/+ +/-/+ -/+/+ +/+/+ Total Low remarks 11 3 2 14 2 1 2 66 101 High remarks 0 0 0 4 0 0 0 100 104 Nucleus 79 0 0 0 0 0 0 0 79 Total 90 3 2 18 2 1 2 166 284 3.2 Diagnostic test accuracy Table 4 shows the estimated sensitivity and specificity of the three tests using the recommended cut-off values for models with and without conditional dependence between tests. All three tests showed high sensitivity and specificity in both models. The 1–12 ELISA and the MFIA had median sensitivities of about 98–99%, while the ApxIV ELISA was slightly lower at around 89–90%. Median specificities were above 96% for all tests. Differences between the two models were minimal, and the estimated covariances between tests were close to zero, indicating negligible dependence between tests and supporting the assumption of conditional independence. Table 4 Posterior estimates (median and 95% posterior credible interval) for the sensitivity, specificity, and conditional covariance (γ) for the 1–12 ELISA, MFIA, and ApxIV ELISA for latent class models with and without conditional independence between tests and minimally informative priors. Parameter Model: without covariance Model: with covariance Median 95% PCI Median 95% PCI Se1-12 ELISA 98.1% [95.3; 99.6] 97.3% [93.7; 99.4] SeMFIA 98.7% [95.9; 99.8] 98.5% [95.1; 99.8] SeApxIV ELISA 89.4% [84.3; 93.3] 88.0% [82.8; 92.2] γSe1-12 ELISA/MFIA 0.000 [-0.001; 0.003] γSe1-12 ELISA/ApxIV -0.001 [-0.005; 0.001] γSe MFIA/ApxIV ELISA 0.000 [-0.004; 0.002] Sp1-12 ELISA 96.9% [91.7; 99.6] 97.2% [90.4; 100.] SpMFIA 98.0% [93.4; 99.9] 97.2% [92.4; 100.] SpApxIV ELISA 97.3% [92.6; 99.5] 99.0% [92.8; 100.] γSp1-12 ELISA/MFIA 0.000 [-0.002; 0.001] γSp1-12 ELISA/ApxIV ELISA 0.000 [-0.001; 0.003] γSp MFIA/ApxIV ELISA 0.000 [-0.001; 0.003] Using the model without covariance, the posterior estimates for the true prevalence [95% PCI] were 83.3% [74.8–90.2] for the herds with the lowest remarks, 99.3% [96.6–100] for the herds with the highest remarks, and 0.9% [0-4.5%] for the genetic nucleus herd. We evaluated the sensitivity and specificity of using multiple tests with serial and parallel interpretations (Table 5 ). In serial interpretation, where all tests must be positive for a positive result, the combination of 1–12 ELISA and MFIA provided the highest sensitivity, while the other combinations showed slightly lower sensitivity but still maintained high specificity. In parallel interpretation, where only one positive test is required for a positive result, all combinations achieved very high sensitivity, with only minor differences in specificity across the pairs. Overall, parallel interpretation improved sensitivity compared to serial interpretation, though at the cost of slightly reduced specificity. Table 5 Sensitivity and specificity (median and 95% posterior credibility interval) of serial and parallel interpretations of different test combinations with 1–12 ELISA, MFIA, and ApxIV ELISA, estimated from the posterior distributions of test properties in the model without covariance. Tests Serial reading Parallel reading Se Sp Se Sp Median 95% PCI Median 95% PCI Median 95% PCI Median 95% PCI 1–12 ELISA + MFIA 96.7% [92.9; 98.9] 99.9% [99.7; 100] 100% [99.9; 100] 94.6% [88.3; 98.6] 1–12 ELISA + ApxIV ELISA 87.6% [82.0; 92.0] 99.9% [99.6; 100] 99.8% [99.4; 100] 93.9% [87.6; 98.0] MFIA + ApxIV ELISA 88.1% [82.5; 92.4] 99.7% [99.7; 100] 99.9% [99.5; 100] 95.0% [89.1; 98.6] To describe the probability that positive and negative test results reflect the true infectious status, positive predictive values (PPV) and negative predictive values (NPV) are shown in Fig. 2 for the 1–12 ELISA, MFIA, and ApxIV ELISA, as individual tests and in serial and parallel of the 1–12 ELISA and MFIA. As expected, the PPV increased with increasing prevalence, while the NPV decreased. Among single tests, the MFIA showed the highest PPV and NPV across most of the prevalence range. For combinations, parallel interpretation of the 1–12 ELISA-MFIA yielded the highest NPV, but the lowest PPV. In contrast, serial interpretation gave the highest PPV and lower NPV relative to the MFIA and 1–12 ELISA. 3.3 Optimizing the cut-off value for the ApxIV ELISA We observed that there were samples close to the recommended cut-off value for the ApxIV ELISA (Fig. 3 ). The ROC curve in Fig. 4 shows the estimated sensitivity and specificity of the test for different cut-off values. We estimated the optimal cut-off from the posterior positive probabilities to be 10, which is lower than the recommended cut-off at 40, assuming that inconclusive results are positive. The dichotomized results with the new cut-off showed higher agreement between the three tests (Table 6 ). Twenty-eight (9.8%) samples showed disagreement between tests with the original cut-off, which was reduced to 16 (5.6%) with the new cut-off. Re-fitting the model using the new cut-off yielded similar results for the estimated Se and Sp of the 1–12 ELISA or MFIA but confirmed a higher median sensitivity of the ApxIV ELISA (Table 7 ). The true prevalence estimates were largely unchanged at 84.3% [76.8–90.8] for the herds with the lowest remarks, 99.3% [96.6–100] for the herds with the highest remarks, and 0.9% [0-4.5%] for the genetic nucleus herd. Table 6 Paired test outcomes for all samples for the 1–12 ELISA, MFIA, and ApxIV ELISA, using different cut-offs for the ApxIV ELISA test. ApxIV ELISA Cut-off Test results for 1–12 ELISA / MFIA / ApxIV ELISA -/-/- +/-/- -/+/- +/+/- -/-/+ +/-/+ -/+/+ +/+/+ Original = 40 90 3 2 18 2 1 2 166 Optimal = 10 88 2 1 4 4 2 3 180 Table 7 Posterior estimates (median and 95% posterior credible interval) for sensitivity and specificity of the 1–12 ELISA, MFIA, and ApxIV ELISA tests using different cut-off values for the ApxIV ELISA. The parameters were estimated from a latent class model assuming conditional independence between tests and minimally informative priors. ApxIV ELISA Cut-off 1–12 ELISA MFIA ApxIV ELISA Se Sp Se Sp Se Sp 40 98.1% [95.3; 99.6] 96.9% [91.7; 99.6] 98.7% [95.9; 99.8] 98.0% [93.4; 99.9] 89.4% [84.3; 93.3] 97.3% [92.6; 99.5] 10 97.9% [95.0; 99.4] 97.4% [92.6; 99.6] 98.4% [95.7; 99.7] 98.5% [94.5; 99.9] 97.3% [94.2; 99.1] 95.2% [89.5; 98.5] 4. Discussion We used a Bayesian latent class model to estimate the diagnostic performance of a microsphere-based multiplex fluorescent immunoassay and two commercial ELISA tests for the serological detection of APP-specific antigens in Norwegian swine herds. The 1–12 ELISA and ApxIV ELISA tests are routinely used by laboratories at the Norwegian Veterinary Institute for monitoring SPF herds, but the test performance has not previously been investigated using field samples from Norway. We found that the ApxIV ELISA had the lowest sensitivity of the three tests at 89.4%, while all three tests showed high specificity. Our results were consistent with estimates from earlier studies and the test manufacturers. Previous studies have documented considerable variation in the sensitivity of the ApxIV-based ELISA, ranging from 74% to 94% using serum samples from pigs with known exposure (Dreyfus et al., 2004 ; Opriessnig et al., 2013 ). Documentation from the manufacturer reports a diagnostic test sensitivity of 83%, which was somewhat lower than our estimate (IDEXX, 2011 ). For the 1–12 ELISA, we found high sensitivity and specificity. Although we did not find estimates using this kit in other studies, the manufacturer stated a diagnostic sensitivity of 86.9% and specificity of 100% (IDvet, 2018 ). To our knowledge, only one previous study has evaluated the performance of the MFIA using LC-LPS to detect APP 1-9-11, 2, 3-6-8-15, 4–7, and 5. Caya et al. ( 2014 ) assessed test accuracy using samples with known APP status based on herd history and LC-LPS ELISA results. They reported relative sensitivities to be from 87.3% to 100% and specificities exceeding 94.6%, depending on the serotype, which agreed with our findings (Caya et al., 2014 ). Nonetheless, direct comparisons across studies should be made cautiously, as differences in study design and sample origin can influence apparent sensitivity and specificity. This study used Bayesian latent class modelling to estimate test characteristics. As discussed by (Toft et al., 2005 ), it is important to evaluate the model assumptions to address sources of potential biases. Our model was a three-test three-population model. The estimated prevalence in the three populations differed markedly, ranging from a median of 0.9% to 99.3%, and we therefore consider the assumption of differing prevalence to be met. All three tests under evaluation detect antibodies in serum samples and are based on the same biological principle. As discussed by (Gardner et al., 2000 ), the tests could therefore be assumed to be dependent. Although our latent state was defined as infection with or previous exposure to APP serotypes 1–12 and 15, the target condition was antibody production in response to APP infection. Any conditional dependence between the tests would be due to potential correlated cross-reactions between the tests (Olsen et al., 2022 ). Adding the covariance for the sensitivity analysis showed only a negligible effect on the estimates of the Se, Sp and true prevalence; therefore, further analysis used the model without covariance. It is important to emphasize that there is a delay in acquiring the antibody response after an infection with APP (Furesz et al., 1997 ), and thus the sensitivity of the tests will be lower than our estimates in a population recently infected. We assumed that the test characteristics were constant across populations, supported by consistent laboratory processing of all samples; however, age-related biological variation and differences in APP exposure due to herd management and geographic location mean we cannot exclude unaccounted for population-level differences in test performance. With only three populations and no population-specific Se and Sp parameters, the model relies on standard assumptions for identifiability. We estimated the optimal cut-off of the ApxIV ELISA using the approach described in Olsen et al. ( 2022 ) to maximize the sum of the sensitivity and specificity. Our results suggest that lowering the cut-off could significantly improve the sensitivity for our samples, without substantially compromising the specificity. However, given the small number of observations near the cut-off in our samples, we suggest a larger and more diverse sample set is needed before recommending a revised cut-off for broader use. Nonetheless, our findings show that lowering the ApxIV ELISA cut-off could improve the sensitivity of the test, which has been suggested by others (Eamens et al., 2012 ; González et al., 2017 ; Opriessnig et al., 2013 ). Several factors can contribute to varying estimates for the sensitivity and specificity of diagnostic tests for APP. These factors include the method of evaluation (e.g., use of imperfect reference standards or incorrectly classified samples), differences in antibody responses (e.g., due to age, time since infection, or maternal antibody levels) (Dreyfus et al., 2004 ; Opriessnig et al., 2013 ; Sjölund et al., 2011 ), the use of experimentally versus naturally infected animals (Gottschalk, 2015 ), and operational differences in sample collection and processing between laboratories. It is also well recognized that test sensitivity and specificity can vary between animal populations for biological reasons that may be difficult to observe (Greiner and Gardner, 2000 ). Consequently, the World Organization for Animal Health (WOAH) recommends validating diagnostic assays in the population where they will be used to ensure that they are fit for purpose (World Organisation for Animal Health, 2021 ). For APP, the presence and distribution of different serotypes can be an important factor contributing to differences in test accuracy. The prevalence of serotypes worldwide is highly varied and the predominant serotypes in European countries differ from those in Asia, Australia, Canada and USA (Soto Perezchica et al., 2023 ). In Norway, serotype 8 is the most frequently detected serotype during clinical outbreaks of acute contagious pleuropneumonia but various other serotypes including serotype 4 and 7 have been detected in conventional herds (Cohen et al., 2020 ). The choice of testing strategy for APP can depend on the surveillance goals, available resources, and the expected herd-level prevalence. In Norway, most testing supports SPF herd monitoring, where the expected prevalence is low and the primary objective is to quickly detect a new introduction. In this context, high sensitivity is important for early detection in the absence of suspicious overt clinical signs. At the same time, high specificity can be desirable to avoid false positives, which can lead to costly confirmatory testing or restrictions on herds. Given that the MFIA and the ELISAs all had high specificity, our findings suggest that the ApxIV ELISA, while routinely used, may be less suitable as a stand-alone screening test than other commercially available ELISAs due to lower sensitivity. However, it has been hypothesized that the sensitivity of the ApxIV ELISA will be high if an APP strain is introduced into a fully naïve herd (Gottschalk, 2015 ). Combining tests can be used to tailor performance to surveillance needs. Parallel testing increases overall sensitivity and improves NPV, which can be useful to rule out infection in high-risk scenarios, such as live-animal trade. We found that combining the 1–12 ELISA and the MFIA in parallel provided the highest sensitivity. Serial testing raises overall specificity and improves PPV, which is particularly valuable in a low-prevalence SPF setting. For example, screening with the 1–12 ELISA and confirming with the MFIA can achieve a high PPV while maintaining good sensitivity. As PPV and NPV depend on prevalence, PPV can be modest in low prevalence even with highly specific assays, whereas NPV remains high. Therefore, interpreting positive results in SPF herds should warrant further investigation, including confirmatory testing and consideration of additional information such as serotype, herd history, and risk. Multiple APP serotypes can be detected with a single ELISA, provided that the assay incorporates polysaccharide-based surface antigens from the relevant serotypes. However, such assays do not allow differentiation between individual serotypes within the same test. In contrast, microsphere-based multiplex fluorescent immunoassays enable the simultaneous and distinct measurement of antibodies to multiple serotypes or serogroups. This approach reduces sample volume, assay time, labor, and inter-assay variability. When serotype-specific data are required, MFIA provides a more efficient and cost-effective alternative to running multiple individual ELISAs. Additionally, MFIAs have demonstrated the potential to detect lower amounts of analyte present in samples than conventional ELISAs (Baker et al., 2012 ; Powell et al., 2013 ; Wagner and Freer, 2009 ). Notably, sera that tested negative by the ApxIV ELISA were found to be positive using an MFIA based on the same antigen (Giménez-Lirola et al., 2014 ). In the present study, the MFIA showed higher sensitivity than both ELISAs, although its performance was largely comparable to that of the 1–12 ELISA. However, it should be emphasized that our evaluation considered MFIA performance at the assay level rather than for individual serogroups, which can obscure potential variation in performance among antigen-specific components. 5. Conclusion This study demonstrates that the MFIA and the 1–12 ELISA provide high sensitivity and specificity for APP in Norwegian swine herds, whereas the ApxIV ELISA shows comparatively lower sensitivity and may be less suitable as a stand-alone screening tool in low-prevalence settings. While adjusting the ApxIV cut-off may improve its sensitivity, further evaluation is needed before recommending changes to routine practice. Overall, our findings support the value of multiplex immunoassays for APP surveillance and highlight the importance of validating diagnostic tests within the populations where they will be applied to ensure accurate and effective disease monitoring. Declarations Animal Ethics statement: Approval from the Norwegian Food Safety Authority was not required, as the study did not involve animal experimentation as defined by the Regulation on the Use of Animals in Experiments (FOR 2015 06 18 761). Blood samples from ‘Små i Ro’ were collected during exsanguination in connection with routine slaughter at a commercial abattoir. Samples from the genetic nucleus herd consisted of pre-existing, stored serum samples collected as part of routine health screening of breeding animals. Declaration of Competing Interest The authors collaborate with Biovet on a separate diagnostic test development project; however, the company had no role in any aspect of the evaluation reported in this paper. All assessments were conducted independently by the authors. The authors declare no additional competing financial or personal interests that could appear to influence the work reported in this paper. Author contributions Katharine R. Dean: Methodology, Software, Data curation, Formal analysis, Visualization, Writing-Original Draft Kari Lybeck: Conceptualization, Resources, Investigation, Writing-Original Draft Ingunn Anita Samdal: Conceptualization, Investigation, Writing - Review & Editing Anniken Jerre Borge: Formal analysis, Writing - Review & Editing Elisabeth Skatvedt Jordal: Conceptualization, Resources, Writing - Review & Editing Carl Andreas Grøntvedt: Conceptualization, Resources, Writing – Review & Editing Sondre Stokke Naadland: Conceptualization, Writing – Review & Editing Kristin Udjus: Investigation, Writing - Review & Editing Irene Haugen: Investigation, Writing - Review & Editing Siv Klevar: Conceptualization, Resources, Project administration, Investigation, Writing - Review & Editing Acknowledgements This study received financial support from the Research Council of Norway and the Research Funding for Agriculture and the Food Industry (FFL/JA) through project number 326686 PreparePig. References Arnal Bernal JL, Gottschalk M, Lacotoure S, Sanz Tejero C, Chacón Pérez G, Martín-Jurado D, Fernández Ros AB (2024) Serotype diversity of Actinobacillus pleuropneumoniae detected by real-time PCR in clinical and subclinical samples from spanish pig farms during 2017–2022. Vet Res 55:165. https://doi.org/10.1186/s13567-024-01419-2 Baker HN, Murphy R, Lopez E, Garcia C (2012) Conversion of a capture ELISA to a Luminex xMAP assay using a multiplex antibody screening method. J Vis Exp 4084. https://doi.org/10.3791/4084 Berger SS, Lauritsen KT, Boas U, Lind P, Andresen LO (2017) Simultaneous detection of antibodies to five Actinobacillus pleuropneumoniae serovars using bead-based multiplex analysis. J Vet Diagn Invest 29:797–804. https://doi.org/10.1177/1040638717719481 Blackall PJ, Klaasen HLBM, Van Den Bosch H, Kuhnert P, Frey J (2002) Proposal of a new serovar of Actinobacillus pleuropneumoniae : serovar 15. Vet Microbiol 84:47–52. https://doi.org/10.1016/S0378-1135(01)00428-X Boekema BKHL, Kamp EM, Smits MA, Smith HE, Stockhofe-Zurwieden N (2004) Both ApxI and ApxII of Actinobacillus pleuropneumoniae serotype 1 are necessary for full virulence. Vet Microbiol 100:17–23. https://doi.org/10.1016/j.vetmic.2003.09.024 Bossé JT, Janson H, Sheehan BJ, Beddek AJ, Rycroft AN, Kroll JS, Langford PR (2002) Actinobacillus pleuropneumoniae : pathobiology and pathogenesis of infection. Microbes Infect 4:225–235. https://doi.org/10.1016/s1286-4579(01)01534-9 Broes A, Martineau G-P, Gottschalk M (2007) Dealing with unexpected Actinobacillus pleuropneumoniae serological results. J Swine Health Prod 15:264–269. https://doi.org/10.54846/jshap/524 Caya I, Bertrand M, Broes A (2014) A multiplexed fluorometric immunoassay (MFIA) for the detection of antibodies to Actinobacillus pleuropneumoniae 1-9-11, 2, 3-6-8-15, 4–7, 5, 10 and 12, in: Proceedings of the American Association of Veterinary Laboratory Diagnosticians Annual Conference. p. 188 Christopher-Hennings J, Araujo KPC, Souza CJH, Fang Y, Lawson S, Nelson EA, Clement T, Dunn M, Lunney JK (2013) Opportunities for bead-based multiplex assays in veterinary diagnostic laboratories. J Vet Diagn Invest 25:671–691. https://doi.org/10.1177/1040638713507256 Cohen LM, Grøntvedt CA, Klem TB, Gulliksen SM, Ranheim B, Nielsen JP, Valheim M, Kielland C (2020) A descriptive study of acute outbreaks of respiratory disease in Norwegian fattening pig herds. Acta Vet Scand 62:35. https://doi.org/10.1186/s13028-020-00529-z Costa G, Oliveira S, Torrison J, Dee S (2011) Evaluation of Actinobacillus pleuropneumoniae diagnostic tests using samples derived from experimentally infected pigs. Vet Microbiol 148:246–251. https://doi.org/10.1016/j.vetmic.2010.08.023 Denwood MJ (2016) runjags: an R package providing interface utilities, model templates, parallel computing methods and additional distributions for MCMC models in JAGS. J Stat Softw 71:1–25. https://doi.org/10.18637/jss.v071.i09 Dreyfus A, Schaller A, Nivollet S, Segers RP, a. M, Kobisch M, Mieli L, Soerensen V, Hüssy D, Miserez R, Zimmermann W, Inderbitzin F, Frey J (2004) Use of recombinant ApxIV in serodiagnosis of Actinobacillus pleuropneumoniae infections, development and prevalidation of the ApxIV ELISA. Vet Microbiol 99:227–238. https://doi.org/10.1016/j.vetmic.2004.01.004 Dubreuil JD, Jacques M, Mittal KR, Gottschalk M (2000) Actinobacillus pleuropneumoniae surface polysaccharides: their role in diagnosis and immunogenicity. Anim Health Res Rev 1:73–93. https://doi.org/10.1017/S1466252300000074 Eamens G, Gonsalves J, Whittington A-M, Turner B (2012) Evaluation of serovar-independent ELISA antigens of Actinobacillus pleuropneumoniae in pigs following vaccination or experimental challenge with respiratory pathogens and natural A. pleuropneumoniae serovar 1 challenge. Aust Vet J 90:490–498. https://doi.org/10.1111/j.1751-0813.2012.01008.x Enøe C, Andersen S, Sørensen V, Willeberg P (2001) Estimation of sensitivity, specificity and predictive values of two serologic tests for the detection of antibodies against Actinobacillus pleuropneumoniae serotype 2 in the absence of a reference test (gold standard). Prev Vet Med 51:227–243. https://doi.org/10.1016/S0167-5877(01)00226-4 Furesz SE, Mallard BA, Bossé JT, Rosendal S, Wilkie BN, MacInnes JI (1997) Antibody- and cell-mediated immune responses of Actinobacillus pleuropneumoniae -infected and bacterin-vaccinated pigs. Infect Immun 65:358–365. https://doi.org/10.1128/iai.65.2.358-365.1997 Gardner IA, Stryhn H, Lind P, Collins MT (2000) Conditional dependence between tests affects the diagnosis and surveillance of animal diseases. Prev Vet Med 45:107–122. https://doi.org/10.1016/S0167-5877(00)00119-7 Giménez-Lirola LG, Jiang Y-H, Sun D, Hoang H, Yoon K-J, Halbur PG, Opriessnig T (2014) Simultaneous detection of antibodies against apx toxins ApxI, ApxII, ApxIII, and ApxIV in pigs with known and unknown Actinobacillus pleuropneumoniae exposure using a multiplexing liquid array platform. Clin Vaccine Immunol 21:85–95. https://doi.org/10.1128/CVI.00451-13 González W, Giménez-Lirola LG, Holmes A, Lizano S, Goodell C, Poonsuk K, Sitthicharoenchai P, Sun Y, Zimmerman J (2017) Detection of Actinobacillus pleuropneumoniae ApxIV toxin antibody in serum and oral fluid specimens from pigs inoculated under experimental conditions. J Vet Res 61:163–171. https://doi.org/10.1515/jvetres-2017-0021 Gottschalk M (2015) The challenge of detecting herds sub-clinically infected with Actinobacillus pleuropneumoniae . Vet J 206:30–38. https://doi.org/10.1016/j.tvjl.2015.06.016 Gottschalk M (2012) Actinobacillus. In: Zimmerman JJ, Karriker LA, Ramirez A, Schwartz KJ, Stevenson GW (eds) Diseases of Swine. Wiley-Blackwell, Ames, IA, pp 653–669 Gottschalk M, Altman E, Charland N, De Lasalle F, Dubreuil JD (1994) Evaluation of a saline boiled extract, capsular polysaccharides and long-chain lipopolysaccharides of Actinobacillus pleuropneumoniae serotype 1 as antigens for the serodiagnosis of swine pleuropneumonia. Vet Microbiol 42:91–104. https://doi.org/10.1016/0378-1135(94)90009-4 Gottschalk M, Altman E, Lacouture S, De Lasalle F, Dubreuil JD (1997) Serodiagnosis of swine pleuropneumonia due to Actinobacillus pleuropneumoniae serotypes 7 and 4 using long-chain lipopolysaccharides. Can J Vet Res 61:62–65 Greiner M, Gardner IA (2000) Epidemiologic issues in the validation of veterinary diagnostic tests. Prev Vet Med 45:3–22. https://doi.org/10.1016/S0167-5877(00)00114-8 Grøntvedt CA, Jordal ES, Valheim M, Urdahl AM, Ellingsen-Dalskau K (2023) Svin. Dyrehelserapporten 2022. Veterinærinstituttet Gulliksen SM, Baustad B, Framstad T, Jørgensen A, Skomsøy A, Kjelvik O, Gjestvang M, Grøntvedt CA, Lium B (2021) Successful eradication of Mycoplasma hyopneumoniae from the Norwegian pig population – 10 years later. Porc Health Manag 7:37. https://doi.org/10.1186/s40813-021-00216-z Hui SL, Walter SD (1980) Estimating the Error Rates of Diagnostic Tests. Biometrics 36:167–171. https://doi.org/10.2307/2530508 IDEXX (2011) APP-Apx IV Ab test validation data report. IDEXX Laboratories Inc IDvet (2018) Internal validation report ID Screen® APP Screening Indirect. Innovative Diagnostics Innovative, Diagnostics (2026) Personal communication via e-mail Klausen J, Ekeroth L, Grøndahl-Hansen J, Andresen LO (2007) An indirect enzyme-linked immunosorbent assay for detection of antibodies to Actinobacillus pleuropneumoniae serovar 7 in pig serum. J Vet Diagn Invest 19:244–249. https://doi.org/10.1177/104063870701900303 Kostoulas P, Nielsen SS, Branscum AJ, Johnson WO, Dendukuri N, Dhand NK, Toft N, Gardner IA (2017) STARD-BLCM: Standards for the Reporting of Diagnostic accuracy studies that use Bayesian Latent Class Models. Prev Vet Med 138:37–47. https://doi.org/10.1016/j.prevetmed.2017.01.006 Kwan W-F, Li Y, Bossé JT, Chiou M-T, Chiu H-J, Langford PR, Mortensen P, Lin C-N (2025) Serovars and antimicrobial resistance profiles of Actinobacillus pleuropneumoniae isolates from clinical-case pigs in Taiwan. BMC Vet Res 21:502. https://doi.org/10.1186/s12917-025-04878-7 Midtveit I (2023) Helse på smågris i Rogaland – ein statusrapport frå prosjektet Små-i-Ro. Svin Midtveit I (2022) Status for smågriskvaliteten i Rogaland. Bondevennen Norsvin SA (2026) Friskere gris med SPF - Norsvin [WWW Document]. URL https://norsvin.no/friskere-gris-med-spf/ (accessed 2.26.26) Olsen A, Nielsen HV, Alban L, Houe H, Jensen TB, Denwood M (2022) Determination of an optimal ELISA cut-off for the diagnosis of Toxoplasma gondii infection in pigs using Bayesian latent class modelling of data from multiple diagnostic tests. Prev Vet Med 201:105606. https://doi.org/10.1016/j.prevetmed.2022.105606 Opriessnig T, Hemann M, Johnson JK, Heinen S, Giménez-Lirola LG, O’Neill KC, Hoang H, Yoon K-J, Gottschalk M, Halbur PG (2013) Evaluation of diagnostic assays for the serological detection of Actinobacillus pleuropneumoniae on samples of known or unknown exposure. J Vet Diagn Invest 25:61–71. https://doi.org/10.1177/1040638712469607 Paulina P, Dawid T (2025) Serotyping and antimicrobial resistance of Actinobacillus pleuropneumoniae isolates from fattening pigs in Poland from 2019 to 2024. BMC Vet Res 21:40. https://doi.org/10.1186/s12917-025-04504-6 Plummer M (2025) rjags: Bayesian graphical models using MCMC Powell RLR, Ouellette I, Lindsay RW, Parks CL, King CR, McDermott AB, Morrow G (2013) A multiplex microsphere-based immunoassay increases the sensitivity of SIV-specific antibody detection in serum samples and mucosal specimens collected from rhesus macaques infected with SIVmac239. BioResearch Open Access 2. https://doi.org/10.1089/biores.2013.0009 . biores.2013.0009 Core Team R (2025) R: a language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria Sassu EL, Bossé JT, Tobias TJ, Gottschalk M, Langford PR, Hennig-Pauka I (2018) Update on Actinobacillus pleuropneumoniae —knowledge, gaps and challenges. Transbound Emerg Dis 65:72–90. https://doi.org/10.1111/tbed.12739 Schaller A, Djordjevic SP, Eamens GJ, Forbes WA, Kuhn R, Kuhnert P, Gottschalk M, Nicolet J, Frey J (2001) Identification and detection of Actinobacillus pleuropneumoniae by PCR based on the gene apxIVA. Vet Microbiol 79:47–62. https://doi.org/10.1016/s0378-1135(00)00345-x Sjölund M, Zoric M, Persson M, Karlsson G, Wallgren P (2011) Disease patterns and immune responses in the offspring to sows with high or low antibody levels to Actinobacillus pleuropneumoniae serotype 2. Res Vet Sci 91:25–31. https://doi.org/10.1016/j.rvsc.2010.07.025 Soto Perezchica MM, Guerrero Barrera AL, Avelar Gonzalez FJ, Quezada Tristan T, Marin M, O (2023) Actinobacillus pleuropneumoniae , surface proteins and virulence: a review. Front Vet Sci 10. https://doi.org/10.3389/fvets.2023.1276712 Stringer OW, Bossé JT, Lacouture S, Gottschalk M, Fodor L, Angen Ø, Velazquez E, Penny P, Lei L, Langford PR, Li Y (2021) Proposal of Actinobacillus pleuropneumoniae serovar 19, and reformulation of previous multiplex PCRs for capsule-specific typing of all known serovars. Vet Microbiol 255:109021. https://doi.org/10.1016/j.vetmic.2021.109021 Stygar AH, Niemi JK, Oliviero C, Laurila T, Heinonen M (2016) Economic value of mitigating Actinobacillus pleuropneumoniae infections in pig fattening herds. Agric Syst 144:113–121. https://doi.org/10.1016/j.agsy.2016.02.005 Tegetmeyer HE, Jones SCP, Langford PR, Baltes N (2008) ISApl 1 , a novel insertion element of Actinobacillus pleuropneumoniae , prevents ApxIV-based serological detection of serotype 7 strain AP76. Vet Microbiol 128:342–353. https://doi.org/10.1016/j.vetmic.2007.10.025 Toft N, Jørgensen E, Højsgaard S (2005) Diagnosing diagnostic tests: evaluating the assumptions underlying the estimation of sensitivity and specificity in the absence of a gold standard. Prev Vet Med 68:19–33. https://doi.org/10.1016/j.prevetmed.2005.01.006 Wagner B, Freer H (2009) Development of a bead-based multiplex assay for simultaneous quantification of cytokines in horses. Vet Immunol Immunopathol 127:242–248. https://doi.org/10.1016/j.vetimm.2008.10.313 Wickham H (2016) ggplot2: Elegant Graphics for Data Analysis. Springer-, New York Wickham H, Averick M, Bryan J, Chang W, McGowan LD, François R, Grolemund G, Hayes A, Henry L, Hester J, Kuhn M, Pedersen TL, Miller E, Bache SM, Müller K, Ooms J, Robinson D, Seidel DP, Spinu V, Takahashi K, Vaughan D, Wilke C, Woo K, Yutani H (2019) Welcome to the tidyverse. J Open Source Softw 4:1686. https://doi.org/10.21105/joss.01686 Wickham H, François R, Henry L, Müller K, Vaughan D (2023) dplyr: A Grammar of Data Manipulation World Organisation for Animal Health (2021) Principles and methods of validation of diagnostic assays for infectious diseases [WWW Document]. Terrestrial Manual. URL https://www.woah.org/fileadmin/Home/fr/Health_standards/tahm/1.01.06_VALIDATION.pdf Additional Declarations The authors declare potential competing interests as follows: The authors collaborate with Biovet on a separate diagnostic test development project; however, the company had no role in any aspect of the evaluation reported in this paper. All assessments were conducted independently by the authors. The authors declare no additional competing financial or personal interests that could appear to influence the work reported in this paper. Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-9608302","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":634138413,"identity":"0ff21f02-3d14-4a4c-b287-72e543d1929e","order_by":0,"name":"Katharine R. Dean","email":"data:image/png;base64,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","orcid":"https://orcid.org/0000-0003-2262-0385","institution":"Norwegian Veterinary Institute","correspondingAuthor":true,"prefix":"","firstName":"Katharine","middleName":"R.","lastName":"Dean","suffix":""},{"id":634138414,"identity":"c39f4d11-9ba6-4c0c-b1fd-786a53557244","order_by":1,"name":"Kari Lybeck","email":"","orcid":"https://orcid.org/0009-0002-2993-2568","institution":"Norwegian Veterinary Institute","correspondingAuthor":false,"prefix":"","firstName":"Kari","middleName":"","lastName":"Lybeck","suffix":""},{"id":634138415,"identity":"ee4851f0-f1c5-480f-a330-0315af814411","order_by":2,"name":"Ingunn Anita Samdal","email":"","orcid":"https://orcid.org/0000-0001-5225-2494","institution":"Norwegian Veterinary Institute","correspondingAuthor":false,"prefix":"","firstName":"Ingunn","middleName":"Anita","lastName":"Samdal","suffix":""},{"id":634138416,"identity":"7637feaf-7f79-4cfe-bab7-cfd44ee691e4","order_by":3,"name":"Anniken Jerre Borge","email":"","orcid":"https://orcid.org/0000-0002-3423-7271","institution":"Norwegian Veterinary Institute","correspondingAuthor":false,"prefix":"","firstName":"Anniken","middleName":"Jerre","lastName":"Borge","suffix":""},{"id":634138417,"identity":"c45d57c3-9565-45da-8714-9c14f66f535c","order_by":4,"name":"Elisabeth Skatvedt Jordal","email":"","orcid":"https://orcid.org/0009-0000-7143-041X","institution":"Norwegian Veterinary Institute","correspondingAuthor":false,"prefix":"","firstName":"Elisabeth","middleName":"Skatvedt","lastName":"Jordal","suffix":""},{"id":634138418,"identity":"4394ce21-ed06-4f2f-b351-b77a73b00e80","order_by":5,"name":"Carl Andreas Grøntvedt","email":"","orcid":"https://orcid.org/0000-0001-6731-4387","institution":"Norwegian Veterinary Institute","correspondingAuthor":false,"prefix":"","firstName":"Carl","middleName":"Andreas","lastName":"Grøntvedt","suffix":""},{"id":634138419,"identity":"11de1dee-cbe3-438d-b426-682c5621e1b4","order_by":6,"name":"Sondre Stokke Naadland","email":"","orcid":"https://orcid.org/0009-0003-7373-7118","institution":"Animalia","correspondingAuthor":false,"prefix":"","firstName":"Sondre","middleName":"Stokke","lastName":"Naadland","suffix":""},{"id":634138420,"identity":"bf4c9b58-f092-4d2a-aad1-ff32ecc2a1f9","order_by":7,"name":"Kristin Udjus","email":"","orcid":"https://orcid.org/0009-0001-1733-490X","institution":"Norwegian Veterinary Institute","correspondingAuthor":false,"prefix":"","firstName":"Kristin","middleName":"","lastName":"Udjus","suffix":""},{"id":634138421,"identity":"13f97c4d-8058-45f2-9ad1-06bef53d80ae","order_by":8,"name":"Irene Haugen","email":"","orcid":"","institution":"Norwegian Veterinary Institute","correspondingAuthor":false,"prefix":"","firstName":"Irene","middleName":"","lastName":"Haugen","suffix":""},{"id":634138422,"identity":"5dde3052-e62e-40f2-b083-615edda61d15","order_by":9,"name":"Siv Klevar","email":"","orcid":"https://orcid.org/0000-0002-0640-1185","institution":"Norwegian Veterinary Institute","correspondingAuthor":false,"prefix":"","firstName":"Siv","middleName":"","lastName":"Klevar","suffix":""}],"badges":[],"createdAt":"2026-05-04 12:20:51","currentVersionCode":1,"declarations":{"humanSubjects":false,"vertebrateSubjects":false,"conflictsOfInterestStatement":true,"humanSubjectEthicalGuidelines":false,"humanSubjectConsent":false,"humanSubjectClinicalTrial":false,"humanSubjectCaseReport":false,"vertebrateSubjectEthicalGuidelines":false},"doi":"10.21203/rs.3.rs-9608302/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-9608302/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":108931225,"identity":"6aa06f6b-1222-497a-a6ec-7ce150decece","added_by":"auto","created_at":"2026-05-11 02:20:46","extension":"jpeg","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":126668,"visible":true,"origin":"","legend":"\u003cp\u003eMap of Norway showing Rogaland county (in blue) where the 20 herds from ‘Små-i-Ro' were located and Vestfold county (in yellow) where the genetic nucleus herd was located.\u003c/p\u003e","description":"","filename":"floatimage1.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-9608302/v1/593fd3acf5d2696fd8cb3f08.jpeg"},{"id":108931226,"identity":"3d95dd84-5412-4958-b523-203fdc8da356","added_by":"auto","created_at":"2026-05-11 02:20:46","extension":"jpeg","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":430846,"visible":true,"origin":"","legend":"\u003cp\u003ePositive predictive value (PPV) and negative predictive values (NPV) of the 1-12 ELISA, MFIA, and ApxIV ELISA single tests and serial and parallel combinations of the 1-12 ELISA and MFIA. The PPV and NPV were calculated from the posterior estimates of the test properties in the model without covariance.\u003c/p\u003e","description":"","filename":"floatimage2.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-9608302/v1/0478ccc326a6a043dba29ea7.jpeg"},{"id":108931228,"identity":"e92d04f4-887f-4047-8108-35eefacbc9e1","added_by":"auto","created_at":"2026-05-11 02:20:47","extension":"jpeg","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":133139,"visible":true,"origin":"","legend":"\u003cp\u003eHistogram showing the distribution of the S/P%-values from the ApxIV ELISA. The red line shows the manufacturer’s recommended cut-off value, when considering inconclusive results as positive.\u003c/p\u003e","description":"","filename":"floatimage3.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-9608302/v1/ccecd54d3b22ae3351754a28.jpeg"},{"id":108931227,"identity":"c28eba0d-3740-4ce7-b240-3dd2980ffe97","added_by":"auto","created_at":"2026-05-11 02:20:47","extension":"jpeg","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":139578,"visible":true,"origin":"","legend":"\u003cp\u003eReceiver operator curve with 95% credible intervals (grey) for the ApxIV ELISA. The asterisk shows the optimal cut-off value at 10.4 with the maximum value for the sum of the sensitivity and specificity.\u003c/p\u003e","description":"","filename":"floatimage4.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-9608302/v1/c94a3eaeb4de714ac474e3a8.jpeg"},{"id":108931241,"identity":"3d0b9aa7-1591-4c26-a2e2-7677986d4571","added_by":"auto","created_at":"2026-05-11 02:20:59","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1431746,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-9608302/v1/7b418a81-1a7f-482e-a258-3ab340ebb3ac.pdf"}],"financialInterests":"The authors declare potential competing interests as follows: The authors collaborate with Biovet on a separate diagnostic test development project; however, the company had no role in any aspect of the evaluation reported in this paper. All assessments were conducted independently by the authors. The authors declare no additional competing financial or personal interests that could appear to influence the work reported in this paper. ","formattedTitle":"\u003cp\u003e\u003cstrong\u003eEvaluation of a multiplex assay and two commercial ELISAs for the serological detection of \u003c/strong\u003e\u003cem\u003e\u003cstrong\u003eActinobacillus pleuropneumoniae \u003c/strong\u003e\u003c/em\u003e\u003cstrong\u003eusing Bayesian latent class modelling\u003c/strong\u003e\u003c/p\u003e","fulltext":[{"header":"Highlights","content":"\u003cp\u003e\u0026bull; MFIA and 1\u0026ndash;12 ELISA showed high sensitivity and specificity for APP surveillance\u003c/p\u003e\u003cp\u003e\u0026bull; ApxIV ELISA had lower sensitivity and may be less suitable for screening\u003c/p\u003e\u003cp\u003e\u0026bull; Lowering the ApxIV cut-off could improve sensitivity without loss of specificity\u003c/p\u003e\u003cp\u003e\u0026bull; APP surveillance requires fit-for-purpose test selection based on objectives\u003c/p\u003e"},{"header":"1. Introduction","content":"\u003cp\u003e \u003cem\u003eActinobacillus pleuropneumoniae\u003c/em\u003e (APP) is a contagious respiratory pathogen in swine and the causative agent of porcine pleuropneumonia (Gottschalk, \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e2012\u003c/span\u003e). The disease is associated with significant economic losses for pig producers from increased mortality, antimicrobial and vaccination costs, and decreased growth performance (Stygar et al., \u003cspan citationid=\"CR49\" class=\"CitationRef\"\u003e2016\u003c/span\u003e). APP is a genetically diverse bacterial species with 19 known serotypes recognized worldwide (Arnal Bernal et al., \u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e2024\u003c/span\u003e; Kwan et al., \u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Paulina and Dawid, \u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e2025\u003c/span\u003e; Stringer et al., \u003cspan citationid=\"CR48\" class=\"CitationRef\"\u003e2021\u003c/span\u003e). Serotype specificity is determined primarily by capsular polysaccharides (CPS) and, in some serotypes, the O chain portion of lipopolysaccharide (LPS) (Blackall et al., \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e2002\u003c/span\u003e; Boekema et al., \u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e2004\u003c/span\u003e; Dubreuil et al., \u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e2000\u003c/span\u003e). Cross-reactive antigenic determinants in the LPS O chain form the basis for serological grouping, notably serotypes 1/9/11, 3/6/8/15, and 4/7 (Gottschalk, \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e2012\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eAPP virulence is closely linked to its set of repeat-in-toxins (RTX) toxins (Boekema et al., \u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e2004\u003c/span\u003e; Boss\u0026eacute; et al., \u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e2002\u003c/span\u003e). Different serotypes produce distinct combinations of Apx toxins: ApxI, ApxII, ApxIII, or mixtures such as ApxI/II and ApxII/III. In contrast, ApxIV is produced exclusively by APP and by all known serotypes, making it a highly specific antigenic marker for the species (Dreyfus et al., \u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e2004\u003c/span\u003e; Schaller et al., \u003cspan citationid=\"CR45\" class=\"CitationRef\"\u003e2001\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eClinically, APP infection results in severe fibrinous and necrotizing pleuropneumonia, causing high morbidity and mortality in acutely affected herds (Sassu et al., \u003cspan citationid=\"CR44\" class=\"CitationRef\"\u003e2018\u003c/span\u003e). Pigs that survive the acute phase\u0026mdash;or that are infected sub-clinically\u0026mdash;can become long‑term asymptomatic carriers, acting as reservoirs for within‑herd transmission (Sassu et al., \u003cspan citationid=\"CR44\" class=\"CitationRef\"\u003e2018\u003c/span\u003e). The presence of such carriers complicates eradication efforts and contributes to the persistence of APP despite prevention and control measures.\u003c/p\u003e \u003cp\u003eSerological surveillance plays a key role in detecting and monitoring APP in sub-clinically infected herds (Gottschalk, \u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e2015\u003c/span\u003e; Sassu et al., \u003cspan citationid=\"CR44\" class=\"CitationRef\"\u003e2018\u003c/span\u003e). Several assays have been developed to identify antibodies against APP antigens, including the complement fixation (CF) test, enzyme-linked immunosorbent assays (ELISAs), and microsphere-based multiplex immunoassays (Berger et al., \u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e2017\u003c/span\u003e; Gottschalk, \u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e2015\u003c/span\u003e). The CF test is no longer recommended due to poor sensitivity and specificity for APP detection (Gottschalk, \u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e2015\u003c/span\u003e). Instead, commercial and in-house ELISAs have become primary diagnostic serological tools. ApxIV-based ELISAs exploit the species-specific and universally expressed ApxIV toxin. However, reduced sensitivity has been reported, potentially due to ISApl1-mediated inhibition of ApxIV expression in certain strains (Eamens et al., \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e2012\u003c/span\u003e; Opriessnig et al., \u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e2013\u003c/span\u003e; Tegetmeyer et al., \u003cspan citationid=\"CR50\" class=\"CitationRef\"\u003e2008\u003c/span\u003e). LPS-based ELISAs detect antibodies targeting serotype-specific LPS O-chain structures (Gottschalk, \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e2012\u003c/span\u003e; Gottschalk et al., \u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e1997\u003c/span\u003e). These assays can detect antibodies to individual serotypes or serogroups, and are considered highly sensitive and specific, but they can be costly when multiple serotypes must be tested (Broes et al., \u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e2007\u003c/span\u003e). Additionally, some ELISAs have been based on CPS or common surface proteins (Gottschalk \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e2012\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eMore recently, microsphere‑based multiplexed fluorometric immunoassays (MFIA) have been developed, allowing simultaneous detection and differentiation of antibodies against multiple APP serotypes within a single well (Berger et al., \u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e2017\u003c/span\u003e; Caya et al., \u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e2014\u003c/span\u003e). These assays use the Luminex Multi-Analyte Profiling of analyte x (xMAP) technology, where fluorescently barcoded microspheres are conjugated to serotype‑specific antigens. Fluorescence is quantified using a flow cytometer, enabling high‑throughput testing. Each bead\u0026rsquo;s identity is resolved by its unique internal dye ratio, while antibody binding is detected through phycoerythrin (PE) fluorescence, measured by a secondary laser, allowing simultaneous quantification of multiple analytes (Berger et al., \u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e2017\u003c/span\u003e; Christopher-Hennings et al., \u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e2013\u003c/span\u003e). Although promising, the multiplex assays for APP have not yet undergone extensive validation using field samples.\u003c/p\u003e \u003cp\u003eDiagnostic test accuracy describes how well a test discriminates between infected and non-infected individuals, commonly measured through sensitivity, specificity, and predictive values. For APP, no perfect reference test exists for detecting antibodies in sub-clinically infected pigs (Sassu et al., \u003cspan citationid=\"CR44\" class=\"CitationRef\"\u003e2018\u003c/span\u003e). Consequently, earlier studies have relied mainly on experimentally infected animals or animals of known status, expert classification, or comparison between multiple imperfect tests (Berger et al., \u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e2017\u003c/span\u003e; Caya et al., \u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e2014\u003c/span\u003e; Costa et al., \u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e2011\u003c/span\u003e; Dreyfus et al., \u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e2004\u003c/span\u003e; Eamens et al., \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e2012\u003c/span\u003e; Gottschalk et al., \u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e1994\u003c/span\u003e; Klausen et al., \u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e2007\u003c/span\u003e; Opriessnig et al., \u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e2013\u003c/span\u003e). Bayesian latent class analysis has been underutilized for estimating the performance of APP diagnostic methods, despite being well-suited for evaluating test performance in the absence of a perfect reference test (En\u0026oslash;e et al., \u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e2001\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eAt the Norwegian Veterinary Institute, the ID Screen APP Screening Indirect (serotypes 1\u0026ndash;12) test from Innovative Diagnostics (Grables, France) is used for routine screening for APP, followed by retesting of positive samples using the IDEXX APP-ApxIV Ab Test from IDEXX Laboratories (Westbrook, Maine, USA). For the export of live animals, industry partners have been advised to screen for APP using IDEXX APP-ApxIV Ab only. Because the ID Screen 1\u0026ndash;12 ELISA and the ApxIV ELISA detect antibodies against different antigens, animals in infected herds may test positive in only one of the two assays (Gottschalk, \u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e2015\u003c/span\u003e). Disagreement between the tests can make interpreting the results challenging.\u003c/p\u003e \u003cp\u003eBuilding on the successful population-wide eradication of \u003cem\u003eMycoplasma hyopneumoniae\u003c/em\u003e (Gulliksen et al., \u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e2021\u003c/span\u003e), the Norwegian pig production sector has placed growing emphasis on improving herd health by encouraging piglet producers and finishers to convert to Specific Pathogen Free (SPF) herds (Norsvin SA, \u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e2026\u003c/span\u003e). These SPF herds are managed under stricter health and biosecurity standards to ensure that they remain free from several major pathogens, including APP. Conversion of a conventional pig herd to SPF status typically involves planned depopulation, thorough cleaning and disinfection, and repopulation with SPF animals after a vacancy period of about three weeks. Certification depends on a structured testing program, and once certified, continuous health monitoring is required, with different protocols for breeding and commercial herds.\u003c/p\u003e \u003cp\u003eThere are currently no international standards for how APP freedom should be established and maintained, which diagnostic tests should be used, and under what conditions (Gottschalk, \u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e2015\u003c/span\u003e; Sassu et al., \u003cspan citationid=\"CR44\" class=\"CitationRef\"\u003e2018\u003c/span\u003e). Consequently, improving knowledge about the diagnostic accuracy of available screening tests is essential, as it directly determines the reliability and economic costs associated with declaring herds APP-free under different sampling strategies. To address this research gap, the aim of this study was to estimate the sensitivity, specificity, and predictive values of a microsphere-based multiplex fluorescent immunoassay and two commercial ELISAs for the detection of antibodies against APP in Norwegian swine herds.\u003c/p\u003e"},{"header":"2. Material and methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003e2.1 Study population and sampling\u003c/h2\u003e \u003cp\u003eThe study was primarily conducted in Rogaland, a county in southwest Norway, shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e. Rogaland lies in the region of Norway with the highest number and density of pig herds (Gr\u0026oslash;ntvedt et al., \u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). Norwegian pig production mainly consists of independent producers that are organized into a hierarchical breeding pyramid, with specialized herds for different production stages. APP is presumed to be prevalent in herds throughout the production pyramid except for herds with approved Specific Pathogen Free (SPF) status, and a previous study has highlighted the importance of APP in porcine acute respiratory disease in Norway (Cohen et al., \u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e2020\u003c/span\u003e).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eSerum samples from 20 sow herds and one genetic nucleus herd were originally collected for other reasons and later re-used for this study. Samples from the sow herds were collected in 2022 as part of the \u0026lsquo;Sm\u0026aring;-i-Ro' project to investigate the association between APP and abattoir recordings (Midtveit, \u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e2023\u003c/span\u003e, \u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). The 20 herds from \u0026lsquo;Sm\u0026aring;-i-Ro' represented the 10 sow herds with the highest and lowest abattoir recordings for pleuritis and pericarditis through routine meat inspection from five slaughterhouses in Rogaland. Samples from a genetic nucleus herd in Vestfold were collected in 2024 as part of routine screening of breeding animals that were 8\u0026ndash;10 weeks old. We assumed that the genetic nucleus herd, the herds with high remarks, and the herds with low remarks had different prevalence for APP and considered them three separate populations for the analysis.\u003c/p\u003e \u003cp\u003eBlood samples from the selected herds were collected from \u003cem\u003ev. jugularis\u003c/em\u003e into serum tubes (without anticoagulant) and transported to the laboratory for centrifugation (2500 x g for 5 min) and further processing. Serum was harvested and frozen at \u0026minus;\u0026thinsp;20\u0026deg;C, until analysis.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec4\" class=\"Section2\"\u003e \u003ch2\u003e2.2 Diagnostic tests\u003c/h2\u003e \u003cdiv id=\"Sec5\" class=\"Section3\"\u003e \u003ch2\u003e2.2.1 Microsphere-based multiplex fluorescent immunoassay (MFIA)\u003c/h2\u003e \u003cp\u003eA multiplex method with seven sets of APP antigens coupled to magnetic beads, named Swinecheck MP APP 1-9-11, 2, 3-6-8-15, 4\u0026ndash;7, 5, 10 \u0026amp; 12 from Biovet (Saint-Hyacinthe, QC, Canada) was tested on the swine serum samples. The magnetic beads are coupled to purified long-chain-LPS antigens from APP serotypes/groups for the detection of IgG antibodies against APP1-9-11, APP 2, APP 3-6-8-15, APP 4\u0026ndash;7, APP 5, APP 10 and APP 12. The multiplex assay was run according to the manufacturer\u0026rsquo;s instructions with minor modifications. Serum samples were diluted 1/100 in the sample buffer and mixed well. A mix of seven magnetic beads coupled with APP antigens (Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e) were vortexed (5s), sonicating (5s) and 50 \u0026micro;L was dispensed in each well on a 96 well plate (Bio-Plex Pro Flat Bottom Plates from BioRad (Hercules, CA, USA)). Then, 50 \u0026micro;L of samples, positive and negative controls were added to dedicated wells and incubated 60 min at 800 rpm in dark on a MixMate shaker (Eppendorf, Hamburg, Germany). The beads on the plate were washed 3x with the supplied wash buffer using Bio-Plex ProWash Station with magnetic separator (Bio-Rad, Hercules, CA, USA) between steps. Primary antibody (100 \u0026micro;L) was added to each well, incubated for 30 min on shaker and then washed. Secondary antibody with R-phycoerythrin (PE) (100\u0026micro;L) was added to the wells, incubated for 30 min on the shaker and then washed again. The plate was analysed by applying 125 \u0026micro;L wash buffer to the wells, resuspended on shaker for 2 min, and read in Bio-Plex 200 (Bio-Rad, Hercules, CA, USA) using Bio-Plex Manager Software version 6.2. The signal was measured as median fluorescence intensity (MFI) of at least 50 beads per bead region, and the doublet discriminator gate was set to 5,000\u0026ndash;25,000. The blank signal was subtracted from the sample signal. The controls were used to calculate the sample-to-positive ratio (S/P) using the following formulae:\u003cdiv id=\"Equa\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equa\" name=\"EquationSource\"\u003e\n$$\\:\\frac{\\text{S}}{\\text{P}}\\:=\\frac{\\text{M}\\text{F}\\text{I}\\text{s}\\text{a}\\text{m}\\text{p}\\text{l}\\text{e}\\:-\\:\\text{M}\\text{F}\\text{I}\\text{n}\\text{e}\\text{g}\\text{a}\\text{t}\\text{i}\\text{v}\\text{e}\\:\\text{c}\\text{o}\\text{n}\\text{t}\\text{r}\\text{o}\\text{l}}{\\text{M}\\text{F}\\text{I}\\text{p}\\text{o}\\text{s}\\text{i}\\text{t}\\text{i}\\text{v}\\text{e}\\:\\text{c}\\text{o}\\text{n}\\text{t}\\text{r}\\text{o}\\text{l}\\:-\\:\\text{M}\\text{F}\\text{I}\\text{n}\\text{e}\\text{g}\\text{a}\\text{t}\\text{i}\\text{v}\\text{e}\\:\\text{c}\\text{o}\\text{n}\\text{t}\\text{r}\\text{o}\\text{l}}\\:$$\u003c/div\u003e\u003c/div\u003e\u003c/p\u003e \u003cp\u003eFor the analysis, inconclusive results were considered positive.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eOverview over antigens coupled to beads and their cut offs\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"5\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003eMicrosphere Mix Constituent\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003eBead Code (Region)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colspan=\"3\" nameend=\"c5\" namest=\"c3\"\u003e \u003cp\u003eS/P\u0026nbsp;cut-offs\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u003cb\u003eNegative\u003c/b\u003e\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003e\u003cb\u003eInconclusive\u003c/b\u003e\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003e\u003cb\u003ePositive\u003c/b\u003e\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAPP 1-9-11\u0026nbsp;antigen\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e51\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e\u0026lt;\u0026nbsp;0.40\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.40\u0026nbsp;\u0026le;\u0026nbsp;S/P\u0026nbsp;\u0026lt;\u0026nbsp;0.55\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e\u0026ge;\u0026nbsp;0.55\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAPP 2 antigen\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e52\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e\u0026lt;\u0026nbsp;0.30\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.30\u0026nbsp;\u0026le;\u0026nbsp;S/P\u0026nbsp;\u0026lt;\u0026nbsp;0.40\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e\u0026ge;\u0026nbsp;0.40\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAPP 3-6-8-15 antigen\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e53\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e\u0026lt;\u0026nbsp;0.30\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.30\u0026nbsp;\u0026le;\u0026nbsp;S/P\u0026nbsp;\u0026lt;\u0026nbsp;0.40\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e\u0026ge;\u0026nbsp;0.40\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAPP 4\u0026ndash;7 antigen\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e54\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e\u0026lt;\u0026nbsp;0.40\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.40\u0026nbsp;\u0026le;\u0026nbsp;S/P\u0026nbsp;\u0026lt;\u0026nbsp;0.50\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e\u0026ge;\u0026nbsp;0.50\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAPP 5 antigen\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e55\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e\u0026lt;\u0026nbsp;0.30\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.30\u0026nbsp;\u0026le;\u0026nbsp;S/P\u0026nbsp;\u0026lt;\u0026nbsp;0.40\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e\u0026ge;\u0026nbsp;0.40\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAPP 10 antigen\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e56\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e\u0026lt;\u0026nbsp;0.30\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.30\u0026nbsp;\u0026le;\u0026nbsp;S/P\u0026nbsp;\u0026lt;\u0026nbsp;0.40\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e\u0026ge;\u0026nbsp;0.40\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eAPP 12 antigen\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e57\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e\u0026lt;\u0026nbsp;0.20\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e0.20\u0026nbsp;\u0026le;\u0026nbsp;S/P\u0026nbsp;\u0026lt;\u0026nbsp;0.30\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e\u0026ge;\u0026nbsp;0.30\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec6\" class=\"Section3\"\u003e \u003ch2\u003e2.2.2 ELISAs\u003c/h2\u003e \u003cp\u003eTwo commercial ELISA kits were used to test the swine sera for comparison. The kits were the ID Screen APP Screening Indirect (serotypes 1 through 12) from Innovative Diagnostics (Grables, France), hereafter called the \u0026lsquo;1\u0026ndash;12 ELISA\u0026rsquo;, and the IDEXX APP-ApxIV Ab Test from IDEXX Laboratories (Westbrook, ME, USA), hereafter called the \u0026lsquo;ApxIV ELISA\u0026rsquo;. The 1\u0026ndash;12 ELISA is designed for the detection of IgG antibodies against APP serotypes 1 through 12 in swine serum, plasma, or meat juice, and uses partially extracted, native polysaccharidic surface antigens derived from the different serotypes (Innovative Diagnostics, \u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e2026\u003c/span\u003e). The ApxIV ELISA, which employs a recombinant ApxIV antigen, is intended for the detection of IgG antibodies against APP in swine serum and plasma. Both kits were run according to the manufacturer\u0026rsquo;s instructions. In both ELISAs samples were tested at 1/10 dilution and absorbances were read at 450 nm. The controls were used to calculate the sample-to-positive ratio (S/P%) using the following formula:\u003cdiv id=\"Equb\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equb\" name=\"EquationSource\"\u003e\n$$\\:\\frac{\\text{S}}{\\text{P}}\\text{%}\\:=\\left(\\frac{\\text{O}\\text{D}\\text{s}\\text{a}\\text{m}\\text{p}\\text{l}\\text{e}\\:-\\:\\text{O}\\text{D}\\text{n}\\text{e}\\text{g}\\text{a}\\text{t}\\text{i}\\text{v}\\text{e}\\:\\text{c}\\text{o}\\text{n}\\text{t}\\text{r}\\text{o}\\text{l}}{\\text{O}\\text{D}\\text{p}\\text{o}\\text{s}\\text{i}\\text{t}\\text{i}\\text{v}\\text{e}\\:\\text{c}\\text{o}\\text{n}\\text{t}\\text{r}\\text{o}\\text{l}\\:-\\:\\text{O}\\text{D}\\text{n}\\text{e}\\text{g}\\text{a}\\text{t}\\text{i}\\text{v}\\text{e}\\:\\text{c}\\text{o}\\text{n}\\text{t}\\text{r}\\text{o}\\text{l}}\\right)\\:\\times\\:100$$\u003c/div\u003e\u003c/div\u003e\u003c/p\u003e \u003cp\u003eRecommended S/P% cut offs for the two ELISAs are shown in Table \u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e. Inconclusive results were considered positive for the evaluation.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab2\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eAPP ELISAs and their cut-offs\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"4\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003eELISA kit\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colspan=\"3\" nameend=\"c4\" namest=\"c2\"\u003e \u003cp\u003eS/P%\u0026nbsp;cut-offs\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u003cb\u003eNegative\u003c/b\u003e\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003e\u003cb\u003eInconclusive\u003c/b\u003e\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003e\u003cb\u003ePositive\u003c/b\u003e\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e1\u0026ndash;12 ELISA\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u0026le;\u0026thinsp;25\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e25\u0026nbsp;\u0026le;\u0026nbsp;S/P%\u0026nbsp;\u0026lt;\u0026nbsp;30\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e\u0026ge; 30\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eApxIV ELISA\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e\u0026lt;\u0026nbsp;40\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e40\u0026nbsp;\u0026le;\u0026nbsp;S/P%\u0026nbsp;\u0026lt;\u0026nbsp;50\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e\u0026ge;\u0026nbsp;50\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003c/div\u003e \u003c/div\u003e \u003cdiv id=\"Sec7\" class=\"Section2\"\u003e \u003ch2\u003e2.3 Bayesian latent class analysis\u003c/h2\u003e \u003cp\u003eWe evaluated the performance of the tests using a Bayesian latent class model for three tests and three populations. The analysis uses a latent class to represent an individual\u0026rsquo;s true status for APP, which is not directly observed in the data without a perfect reference test. We defined the latent condition as infection with or previous exposure to APP serotypes 1\u0026ndash;12 and 15. We grouped the herds to form three populations consisting of those with the highest remarks, the lowest remarks, and the genetic nucleus herd, based on the assumption that they would have differing APP prevalence. We specified a Bayesian latent class model based on the Hui-Walter framework (Hui and Walter, \u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e1980\u003c/span\u003e). We assumed that the test characteristics were constant across populations and that the tests were conditionally independent given the infection status.\u003c/p\u003e \u003cp\u003eFor each population, \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:k\\)\u003c/span\u003e\u003c/span\u003e, we summarized the test results as a vector, \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{y}_{k}\\)\u003c/span\u003e\u003c/span\u003e, with the number of individuals in each possible combination of test outcomes. We used a multinomial distribution to model the observed data as follows:\u003cdiv id=\"Equc\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equc\" name=\"EquationSource\"\u003e\n$$\\:\\left.{y}_{k}\\right|{\\pi\\:}_{k}{Se}_{1},{Se}_{2},\\:{Se}_{3,}{Sp}_{1},{Sp}_{2},{Sp}_{3}\u0026sim;Multinomial\\left({p}_{k},\\:{n}_{k}\\right),\\:k=\\text{1,2},3$$\u003c/div\u003e\u003c/div\u003e\u003cdiv id=\"Equd\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equd\" name=\"EquationSource\"\u003e\n$$\\:{p}_{000k}=\\:{\\pi\\:}_{k}(1-{Se}_{1})(1-{Se}_{2})(1-{Se}_{3})+(1-{\\pi\\:}_{k}){Sp}_{1}{Sp}_{2}{Sp}_{3}$$\u003c/div\u003e\u003c/div\u003e\u003cdiv id=\"Eque\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Eque\" name=\"EquationSource\"\u003e\n$$\\:{p}_{100k}=\\:{\\pi\\:}_{k}{Se}_{1}(1-{Se}_{2})(1-{Se}_{3})+(1-{\\pi\\:}_{k})(1-{Sp}_{1}){Sp}_{2}{Sp}_{3}$$\u003c/div\u003e\u003c/div\u003e\u003cdiv id=\"Equf\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equf\" name=\"EquationSource\"\u003e\n$$\\:{p}_{010k}=\\:{\\pi\\:}_{k}(1-{Se}_{1}){Se}_{2}(1-{Se}_{3})+(1-{\\pi\\:}_{k}){Sp}_{1}(1-{Sp}_{2}){Sp}_{3}$$\u003c/div\u003e\u003c/div\u003e\u003cdiv id=\"Equg\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equg\" name=\"EquationSource\"\u003e\n$$\\:{p}_{110k}=\\:{\\pi\\:}_{k}{Se}_{1}{Se}_{2}(1-{Se}_{3})+(1-{\\pi\\:}_{k})(1-{Sp}_{1})(1-{Sp}_{2}){Sp}_{3}$$\u003c/div\u003e\u003c/div\u003e\u003cdiv id=\"Equh\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equh\" name=\"EquationSource\"\u003e\n$$\\:{p}_{001k}=\\:{\\pi\\:}_{k}(1-{Se}_{1})(1-{Se}_{2}){Se}_{3}+(1-{\\pi\\:}_{k}){Sp}_{1}{Sp}_{2}(1-{Sp}_{3})$$\u003c/div\u003e\u003c/div\u003e\u003cdiv id=\"Equi\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equi\" name=\"EquationSource\"\u003e\n$$\\:{p}_{101k}=\\:{\\pi\\:}_{k}{Se}_{1}(1-{Se}_{2}){Se}_{3}+(1-{\\pi\\:}_{k})(1-{Sp}_{1}){Sp}_{2}{(1-Sp}_{3})$$\u003c/div\u003e\u003c/div\u003e\u003cdiv id=\"Equj\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equj\" name=\"EquationSource\"\u003e\n$$\\:{p}_{011k}=\\:{\\pi\\:}_{k}(1-{Se}_{1}){Se}_{2}{Se}_{3}+(1-{\\pi\\:}_{k}){Sp}_{1}(1-{Sp}_{2}){(1-Sp}_{3})$$\u003c/div\u003e\u003c/div\u003e\u003cdiv id=\"Equk\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equk\" name=\"EquationSource\"\u003e\n$$\\:{p}_{111k}=\\:{\\pi\\:}_{k}{Se}_{1}{Se}_{2}{Se}_{3}+(1-{\\pi\\:}_{k})(1-{Sp}_{1}){(1-Sp}_{2}\\left)\\right(1-{Sp}_{3})$$\u003c/div\u003e\u003c/div\u003e\u003c/p\u003e \u003cp\u003ewhere, \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{\\pi\\:}_{k}\\)\u003c/span\u003e\u003c/span\u003e, is the prevalence of APP in each population, and \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:Se\\)\u003c/span\u003e\u003c/span\u003e and \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:Sp\\)\u003c/span\u003e\u003c/span\u003e are the sensitivity and specificity of each test. We used Bayesian analysis to estimate the unknown parameters, which were the sensitivity and specificity of each test, as well as the true prevalence in the populations. Since we lacked knowledge about the true values of these parameters, we assigned minimally informative \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:Beta\\left(\\text{1,1}\\right)\\)\u003c/span\u003e\u003c/span\u003e distributions as priors.\u003c/p\u003e \u003cp\u003eWe estimated the posterior distributions by running two independent Monte-Carlo Markov chains (MCMC) with dispersed starting values for a total of 200,500 iterations, including a 5,000 burn-in and 1/20 thinning, for a total of 20,000 posterior samples per chain. The effective sample size was above 1000 for all variables. We evaluated model convergence by visually inspecting the MCMC trace plots and checking that the Gelman\u0026ndash;Rubin potential scale reduction factor was below 1.05 for all parameters.\u003c/p\u003e \u003cp\u003eWe calculated the mean and median values for the estimated parameters and the 95% credible intervals from the posterior samples (PCI). We used the samples from the MCMC chains to calculate the sensitivity and specificity of the serial and parallel interpretations of the different paired test combinations and the positive and negative predictive values.\u003c/p\u003e \u003cp\u003eWe implemented the model in JAGS version 4.3.1 (Plummer, \u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e2025\u003c/span\u003e) and R version 4.4.3 (R Core Team, \u003cspan citationid=\"CR43\" class=\"CitationRef\"\u003e2025\u003c/span\u003e) using the rjags (Plummer, \u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e2025\u003c/span\u003e), runjags (Denwood, \u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e2016\u003c/span\u003e), dplyr (Wickham et al., \u003cspan citationid=\"CR55\" class=\"CitationRef\"\u003e2023\u003c/span\u003e), tidyr (Wickham et al., \u003cspan citationid=\"CR54\" class=\"CitationRef\"\u003e2019\u003c/span\u003e), and ggplot2 (Wickham, \u003cspan citationid=\"CR53\" class=\"CitationRef\"\u003e2016\u003c/span\u003e). The software code and data for reproducing the results is available on Github: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://github.com/NorwegianVeterinaryInstitute/App_test_evaluation\u003c/span\u003e\u003cspan address=\"https://github.com/NorwegianVeterinaryInstitute/App_test_evaluation\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. The manuscript was prepared using the \u0026ldquo;STARD-BLCM: Standards for the reporting of diagnostic accuracy studies that use Bayesian Latent class models\u0026rdquo; checklist (Kostoulas et al., \u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e2017\u003c/span\u003e).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec8\" class=\"Section2\"\u003e \u003ch2\u003e2.4 Sensitivity analysis\u003c/h2\u003e \u003cp\u003eAs all three tests measured IgG antibodies towards APP, we investigated the assumption of independence between tests as a sensitivity analysis. We compared an alternative model with conditional dependence between tests to the original model without conditional dependence. We used minimally informative \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:Beta\\left(\\text{2,1}\\right)\\)\u003c/span\u003e\u003c/span\u003e distributions as priors for the sensitivity and specificity, and \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:Beta\\left(\\text{1,1}\\right)\\)\u003c/span\u003e\u003c/span\u003e for the prevalence.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec9\" class=\"Section2\"\u003e \u003ch2\u003e2.5 Optimizing the cut-off values for ApxIV ELISA\u003c/h2\u003e \u003cp\u003ePrevious studies have found lower sensitivity for the ApxIV ELISA, suggesting that the manufacturer\u0026rsquo;s recommended cut-off is too high (Eamens et al., \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e2012\u003c/span\u003e; Gonz\u0026aacute;lez et al., \u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e2017\u003c/span\u003e; Opriessnig et al., \u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e2013\u003c/span\u003e). Therefore, we estimated the optimal cut-off value for the ApxIV ELISA for our samples using the posterior probability as described in (Olsen et al., \u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). Following Olsen (2022), we defined the optimal cut-off value as the maximum sum of the sensitivity and specificity. To explore the performance of the test across a range of cut-off values, we constructed a receiver operator curve (ROC) with 95% PCI. We reran the three test-three population model without covariance using the new cut-off value for the ApxIV ELISA test.\u003c/p\u003e \u003c/div\u003e"},{"header":"3. Results","content":"\u003cdiv id=\"Sec11\" class=\"Section2\"\u003e \u003ch2\u003e3.1 Descriptive statistics\u003c/h2\u003e \u003cp\u003eWe tested a total of 284 samples with the three tests: 1\u0026ndash;12 ELISA, MFIA, ApxIV ELISA. When using the recommended cut-off values, 188 (66.2%) samples were positive in the 1\u0026ndash;12 ELISA, 188 (66.2%) were positive in the MFIA, and 171 (62.3%) were positive in the ApxIV ELISA. For 256 (90.1%) samples, there was agreement between all three tests, whereas for 28 (9.9%) samples, the results differed. Table\u0026nbsp;\u003cspan refid=\"Tab3\" class=\"InternalRef\"\u003e3\u003c/span\u003e shows the paired test outcomes based on the recommended cut-off values for each test and subpopulation.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab3\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 3\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003ePaired test outcomes for each subpopulation with the 1\u0026ndash;12 ELISA, MFIA, and ApxIV ELISA, using the manufacturers recommended cut-off values.\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"10\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c7\" colnum=\"7\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c8\" colnum=\"8\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c9\" colnum=\"9\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c10\" colnum=\"10\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003ePopulation\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colspan=\"9\" nameend=\"c10\" namest=\"c2\"\u003e \u003cp\u003eTest results for 1\u0026ndash;12 ELISA / MFIA / ApxIV ELISA\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003e-/-/-\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003e+/-/-\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003e-/+/-\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003e+/+/-\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003e-/-/+\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c7\"\u003e \u003cp\u003e+/-/+\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c8\"\u003e \u003cp\u003e-/+/+\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c9\"\u003e \u003cp\u003e+/+/+\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c10\"\u003e \u003cp\u003eTotal\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eLow remarks\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e11\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e14\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e66\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c10\"\u003e \u003cp\u003e101\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eHigh remarks\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e100\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c10\"\u003e \u003cp\u003e104\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNucleus\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e79\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c10\"\u003e \u003cp\u003e79\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e\u003cb\u003eTotal\u003c/b\u003e\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e90\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e18\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e166\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c10\"\u003e \u003cp\u003e284\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec12\" class=\"Section2\"\u003e \u003ch2\u003e3.2 Diagnostic test accuracy\u003c/h2\u003e \u003cp\u003eTable\u0026nbsp;\u003cspan refid=\"Tab4\" class=\"InternalRef\"\u003e4\u003c/span\u003e shows the estimated sensitivity and specificity of the three tests using the recommended cut-off values for models with and without conditional dependence between tests. All three tests showed high sensitivity and specificity in both models. The 1\u0026ndash;12 ELISA and the MFIA had median sensitivities of about 98\u0026ndash;99%, while the ApxIV ELISA was slightly lower at around 89\u0026ndash;90%. Median specificities were above 96% for all tests. Differences between the two models were minimal, and the estimated covariances between tests were close to zero, indicating negligible dependence between tests and supporting the assumption of conditional independence.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab4\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 4\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003ePosterior estimates (median and 95% posterior credible interval) for the sensitivity, specificity, and conditional covariance (γ) for the 1\u0026ndash;12 ELISA, MFIA, and ApxIV ELISA for latent class models with and without conditional independence between tests and minimally informative priors.\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"5\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003eParameter\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colspan=\"2\" nameend=\"c3\" namest=\"c2\"\u003e \u003cp\u003eModel: without covariance\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colspan=\"2\" nameend=\"c5\" namest=\"c4\"\u003e \u003cp\u003eModel: with covariance\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eMedian\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003e95% PCI\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eMedian\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003e95% PCI\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSe1-12 ELISA\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e98.1%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e[95.3; 99.6]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e97.3%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e[93.7; 99.4]\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSeMFIA\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e98.7%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e[95.9; 99.8]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e98.5%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e[95.1; 99.8]\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSeApxIV ELISA\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e89.4%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e[84.3; 93.3]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e88.0%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e[82.8; 92.2]\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eγSe1-12 ELISA/MFIA\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.000\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e[-0.001; 0.003]\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eγSe1-12 ELISA/ApxIV\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e-0.001\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e[-0.005; 0.001]\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eγSe MFIA/ApxIV ELISA\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.000\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e[-0.004; 0.002]\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSp1-12 ELISA\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e96.9%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e[91.7; 99.6]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e97.2%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e[90.4; 100.]\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSpMFIA\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e98.0%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e[93.4; 99.9]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e97.2%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e[92.4; 100.]\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSpApxIV ELISA\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e97.3%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e[92.6; 99.5]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e99.0%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e[92.8; 100.]\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eγSp1-12 ELISA/MFIA\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.000\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e[-0.002; 0.001]\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eγSp1-12 ELISA/ApxIV ELISA\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.000\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e[-0.001; 0.003]\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eγSp MFIA/ApxIV ELISA\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e\u0026nbsp;\u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.000\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e[-0.001; 0.003]\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eUsing the model without covariance, the posterior estimates for the true prevalence [95% PCI] were 83.3% [74.8\u0026ndash;90.2] for the herds with the lowest remarks, 99.3% [96.6\u0026ndash;100] for the herds with the highest remarks, and 0.9% [0-4.5%] for the genetic nucleus herd.\u003c/p\u003e \u003cp\u003eWe evaluated the sensitivity and specificity of using multiple tests with serial and parallel interpretations (Table\u0026nbsp;\u003cspan refid=\"Tab5\" class=\"InternalRef\"\u003e5\u003c/span\u003e). In serial interpretation, where all tests must be positive for a positive result, the combination of 1\u0026ndash;12 ELISA and MFIA provided the highest sensitivity, while the other combinations showed slightly lower sensitivity but still maintained high specificity. In parallel interpretation, where only one positive test is required for a positive result, all combinations achieved very high sensitivity, with only minor differences in specificity across the pairs. Overall, parallel interpretation improved sensitivity compared to serial interpretation, though at the cost of slightly reduced specificity.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab5\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 5\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eSensitivity and specificity (median and 95% posterior credibility interval) of serial and parallel interpretations of different test combinations with 1\u0026ndash;12 ELISA, MFIA, and ApxIV ELISA, estimated from the posterior distributions of test properties in the model without covariance.\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"10\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c7\" colnum=\"7\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c8\" colnum=\"8\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c9\" colnum=\"9\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c10\" colnum=\"10\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003eTests\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colspan=\"5\" nameend=\"c6\" namest=\"c2\"\u003e \u003cp\u003eSerial reading\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colspan=\"4\" nameend=\"c10\" namest=\"c7\"\u003e \u003cp\u003eParallel reading\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003ctr\u003e \u003cth align=\"left\" colspan=\"3\" nameend=\"c4\" namest=\"c2\"\u003e \u003cp\u003eSe\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colspan=\"2\" nameend=\"c6\" namest=\"c5\"\u003e \u003cp\u003eSp\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colspan=\"2\" nameend=\"c8\" namest=\"c7\"\u003e \u003cp\u003eSe\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colspan=\"2\" nameend=\"c10\" namest=\"c9\"\u003e \u003cp\u003eSp\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eMedian\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003e95% PCI\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colspan=\"2\" nameend=\"c5\" namest=\"c4\"\u003e \u003cp\u003eMedian\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003e95% PCI\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c7\"\u003e \u003cp\u003eMedian\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c8\"\u003e \u003cp\u003e95% PCI\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c9\"\u003e \u003cp\u003eMedian\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c10\"\u003e \u003cp\u003e95% PCI\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e1\u0026ndash;12 ELISA\u0026thinsp;+\u0026thinsp;MFIA\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e96.7%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e[92.9; 98.9]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c5\" namest=\"c4\"\u003e \u003cp\u003e99.9%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e[99.7; 100]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e100%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e[99.9; 100]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e94.6%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e \u003cp\u003e[88.3; 98.6]\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e1\u0026ndash;12 ELISA + ApxIV ELISA\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e87.6%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e[82.0; 92.0]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c5\" namest=\"c4\"\u003e \u003cp\u003e99.9%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e[99.6; 100]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e99.8%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e[99.4; 100]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e93.9%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e \u003cp\u003e[87.6; 98.0]\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eMFIA\u0026thinsp;+\u0026thinsp;ApxIV ELISA\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e88.1%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e[82.5;\u003c/p\u003e \u003cp\u003e92.4]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c5\" namest=\"c4\"\u003e \u003cp\u003e99.7%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e[99.7; 100]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e99.9%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e[99.5; 100]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e95.0%\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c10\"\u003e \u003cp\u003e[89.1; 98.6]\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eTo describe the probability that positive and negative test results reflect the true infectious status, positive predictive values (PPV) and negative predictive values (NPV) are shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e for the 1\u0026ndash;12 ELISA, MFIA, and ApxIV ELISA, as individual tests and in serial and parallel of the 1\u0026ndash;12 ELISA and MFIA. As expected, the PPV increased with increasing prevalence, while the NPV decreased. Among single tests, the MFIA showed the highest PPV and NPV across most of the prevalence range. For combinations, parallel interpretation of the 1\u0026ndash;12 ELISA-MFIA yielded the highest NPV, but the lowest PPV. In contrast, serial interpretation gave the highest PPV and lower NPV relative to the MFIA and 1\u0026ndash;12 ELISA.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec13\" class=\"Section2\"\u003e \u003ch2\u003e3.3 Optimizing the cut-off value for the ApxIV ELISA\u003c/h2\u003e \u003cp\u003eWe observed that there were samples close to the recommended cut-off value for the ApxIV ELISA (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e). The ROC curve in Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e shows the estimated sensitivity and specificity of the test for different cut-off values. We estimated the optimal cut-off from the posterior positive probabilities to be 10, which is lower than the recommended cut-off at 40, assuming that inconclusive results are positive.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eThe dichotomized results with the new cut-off showed higher agreement between the three tests (Table\u0026nbsp;\u003cspan refid=\"Tab6\" class=\"InternalRef\"\u003e6\u003c/span\u003e). Twenty-eight (9.8%) samples showed disagreement between tests with the original cut-off, which was reduced to 16 (5.6%) with the new cut-off. Re-fitting the model using the new cut-off yielded similar results for the estimated Se and Sp of the 1\u0026ndash;12 ELISA or MFIA but confirmed a higher median sensitivity of the ApxIV ELISA (Table\u0026nbsp;\u003cspan refid=\"Tab7\" class=\"InternalRef\"\u003e7\u003c/span\u003e). The true prevalence estimates were largely unchanged at 84.3% [76.8\u0026ndash;90.8] for the herds with the lowest remarks, 99.3% [96.6\u0026ndash;100] for the herds with the highest remarks, and 0.9% [0-4.5%] for the genetic nucleus herd.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab6\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 6\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003ePaired test outcomes for all samples for the 1\u0026ndash;12 ELISA, MFIA, and ApxIV ELISA, using different cut-offs for the ApxIV ELISA test.\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"9\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c7\" colnum=\"7\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c8\" colnum=\"8\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c9\" colnum=\"9\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003eApxIV ELISA Cut-off\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colspan=\"8\" nameend=\"c9\" namest=\"c2\"\u003e \u003cp\u003eTest results for 1\u0026ndash;12 ELISA / MFIA / ApxIV ELISA\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003e-/-/-\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003e+/-/-\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003e-/+/-\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003e+/+/-\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003e-/-/+\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c7\"\u003e \u003cp\u003e+/-/+\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c8\"\u003e \u003cp\u003e-/+/+\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c9\"\u003e \u003cp\u003e+/+/+\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eOriginal\u0026thinsp;=\u0026thinsp;40\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e90\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e18\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e166\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eOptimal\u0026thinsp;=\u0026thinsp;10\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e88\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e4\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c8\"\u003e \u003cp\u003e3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c9\"\u003e \u003cp\u003e180\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab7\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 7\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003ePosterior estimates (median and 95% posterior credible interval) for sensitivity and specificity of the 1\u0026ndash;12 ELISA, MFIA, and ApxIV ELISA tests using different cut-off values for the ApxIV ELISA. The parameters were estimated from a latent class model assuming conditional independence between tests and minimally informative priors.\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"7\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c7\" colnum=\"7\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\" morerows=\"1\" rowspan=\"2\"\u003e \u003cp\u003eApxIV ELISA Cut-off\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colspan=\"2\" nameend=\"c3\" namest=\"c2\"\u003e \u003cp\u003e1\u0026ndash;12 ELISA\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colspan=\"2\" nameend=\"c5\" namest=\"c4\"\u003e \u003cp\u003eMFIA\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colspan=\"2\" nameend=\"c7\" namest=\"c6\"\u003e \u003cp\u003eApxIV ELISA\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eSe\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eSp\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eSe\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eSp\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003eSe\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c7\"\u003e \u003cp\u003eSp\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e40\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e98.1% [95.3; 99.6]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e96.9% [91.7; 99.6]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e98.7% [95.9; 99.8]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e98.0% [93.4; 99.9]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e89.4% [84.3; 93.3]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e97.3% [92.6; 99.5]\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e10\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e97.9% [95.0; 99.4]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e97.4% [92.6; 99.6]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e98.4% [95.7; 99.7]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e98.5% [94.5; 99.9]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c6\"\u003e \u003cp\u003e97.3% [94.2; 99.1]\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c7\"\u003e \u003cp\u003e95.2% [89.5; 98.5]\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003c/div\u003e"},{"header":"4. Discussion","content":"\u003cp\u003eWe used a Bayesian latent class model to estimate the diagnostic performance of a microsphere-based multiplex fluorescent immunoassay and two commercial ELISA tests for the serological detection of APP-specific antigens in Norwegian swine herds. The 1\u0026ndash;12 ELISA and ApxIV ELISA tests are routinely used by laboratories at the Norwegian Veterinary Institute for monitoring SPF herds, but the test performance has not previously been investigated using field samples from Norway. We found that the ApxIV ELISA had the lowest sensitivity of the three tests at 89.4%, while all three tests showed high specificity.\u003c/p\u003e \u003cp\u003eOur results were consistent with estimates from earlier studies and the test manufacturers. Previous studies have documented considerable variation in the sensitivity of the ApxIV-based ELISA, ranging from 74% to 94% using serum samples from pigs with known exposure (Dreyfus et al., \u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e2004\u003c/span\u003e; Opriessnig et al., \u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e2013\u003c/span\u003e). Documentation from the manufacturer reports a diagnostic test sensitivity of 83%, which was somewhat lower than our estimate (IDEXX, \u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e2011\u003c/span\u003e). For the 1\u0026ndash;12 ELISA, we found high sensitivity and specificity. Although we did not find estimates using this kit in other studies, the manufacturer stated a diagnostic sensitivity of 86.9% and specificity of 100% (IDvet, \u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e2018\u003c/span\u003e). To our knowledge, only one previous study has evaluated the performance of the MFIA using LC-LPS to detect APP 1-9-11, 2, 3-6-8-15, 4\u0026ndash;7, and 5. Caya et al. (\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e2014\u003c/span\u003e) assessed test accuracy using samples with known APP status based on herd history and LC-LPS ELISA results. They reported relative sensitivities to be from 87.3% to 100% and specificities exceeding 94.6%, depending on the serotype, which agreed with our findings (Caya et al., \u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e2014\u003c/span\u003e). Nonetheless, direct comparisons across studies should be made cautiously, as differences in study design and sample origin can influence apparent sensitivity and specificity.\u003c/p\u003e \u003cp\u003eThis study used Bayesian latent class modelling to estimate test characteristics. As discussed by (Toft et al., \u003cspan citationid=\"CR51\" class=\"CitationRef\"\u003e2005\u003c/span\u003e), it is important to evaluate the model assumptions to address sources of potential biases. Our model was a three-test three-population model. The estimated prevalence in the three populations differed markedly, ranging from a median of 0.9% to 99.3%, and we therefore consider the assumption of differing prevalence to be met. All three tests under evaluation detect antibodies in serum samples and are based on the same biological principle. As discussed by (Gardner et al., \u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e2000\u003c/span\u003e), the tests could therefore be assumed to be dependent. Although our latent state was defined as infection with or previous exposure to APP serotypes 1\u0026ndash;12 and 15, the target condition was antibody production in response to APP infection. Any conditional dependence between the tests would be due to potential correlated cross-reactions between the tests (Olsen et al., \u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e2022\u003c/span\u003e). Adding the covariance for the sensitivity analysis showed only a negligible effect on the estimates of the Se, Sp and true prevalence; therefore, further analysis used the model without covariance. It is important to emphasize that there is a delay in acquiring the antibody response after an infection with APP (Furesz et al., \u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e1997\u003c/span\u003e), and thus the sensitivity of the tests will be lower than our estimates in a population recently infected. We assumed that the test characteristics were constant across populations, supported by consistent laboratory processing of all samples; however, age-related biological variation and differences in APP exposure due to herd management and geographic location mean we cannot exclude unaccounted for population-level differences in test performance. With only three populations and no population-specific Se and Sp parameters, the model relies on standard assumptions for identifiability.\u003c/p\u003e \u003cp\u003eWe estimated the optimal cut-off of the ApxIV ELISA using the approach described in Olsen et al. (\u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e2022\u003c/span\u003e) to maximize the sum of the sensitivity and specificity. Our results suggest that lowering the cut-off could significantly improve the sensitivity for our samples, without substantially compromising the specificity. However, given the small number of observations near the cut-off in our samples, we suggest a larger and more diverse sample set is needed before recommending a revised cut-off for broader use. Nonetheless, our findings show that lowering the ApxIV ELISA cut-off could improve the sensitivity of the test, which has been suggested by others (Eamens et al., \u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e2012\u003c/span\u003e; Gonz\u0026aacute;lez et al., \u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e2017\u003c/span\u003e; Opriessnig et al., \u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e2013\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eSeveral factors can contribute to varying estimates for the sensitivity and specificity of diagnostic tests for APP. These factors include the method of evaluation (e.g., use of imperfect reference standards or incorrectly classified samples), differences in antibody responses (e.g., due to age, time since infection, or maternal antibody levels) (Dreyfus et al., \u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e2004\u003c/span\u003e; Opriessnig et al., \u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e2013\u003c/span\u003e; Sj\u0026ouml;lund et al., \u003cspan citationid=\"CR46\" class=\"CitationRef\"\u003e2011\u003c/span\u003e), the use of experimentally versus naturally infected animals (Gottschalk, \u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e2015\u003c/span\u003e), and operational differences in sample collection and processing between laboratories. It is also well recognized that test sensitivity and specificity can vary between animal populations for biological reasons that may be difficult to observe (Greiner and Gardner, \u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e2000\u003c/span\u003e). Consequently, the World Organization for Animal Health (WOAH) recommends validating diagnostic assays in the population where they will be used to ensure that they are fit for purpose (World Organisation for Animal Health, \u003cspan citationid=\"CR56\" class=\"CitationRef\"\u003e2021\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eFor APP, the presence and distribution of different serotypes can be an important factor contributing to differences in test accuracy. The prevalence of serotypes worldwide is highly varied and the predominant serotypes in European countries differ from those in Asia, Australia, Canada and USA (Soto Perezchica et al., \u003cspan citationid=\"CR47\" class=\"CitationRef\"\u003e2023\u003c/span\u003e). In Norway, serotype 8 is the most frequently detected serotype during clinical outbreaks of acute contagious pleuropneumonia but various other serotypes including serotype 4 and 7 have been detected in conventional herds (Cohen et al., \u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e2020\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eThe choice of testing strategy for APP can depend on the surveillance goals, available resources, and the expected herd-level prevalence. In Norway, most testing supports SPF herd monitoring, where the expected prevalence is low and the primary objective is to quickly detect a new introduction. In this context, high sensitivity is important for early detection in the absence of suspicious overt clinical signs. At the same time, high specificity can be desirable to avoid false positives, which can lead to costly confirmatory testing or restrictions on herds. Given that the MFIA and the ELISAs all had high specificity, our findings suggest that the ApxIV ELISA, while routinely used, may be less suitable as a stand-alone screening test than other commercially available ELISAs due to lower sensitivity. However, it has been hypothesized that the sensitivity of the ApxIV ELISA will be high if an APP strain is introduced into a fully na\u0026iuml;ve herd (Gottschalk, \u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e2015\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eCombining tests can be used to tailor performance to surveillance needs. Parallel testing increases overall sensitivity and improves NPV, which can be useful to rule out infection in high-risk scenarios, such as live-animal trade. We found that combining the 1\u0026ndash;12 ELISA and the MFIA in parallel provided the highest sensitivity. Serial testing raises overall specificity and improves PPV, which is particularly valuable in a low-prevalence SPF setting. For example, screening with the 1\u0026ndash;12 ELISA and confirming with the MFIA can achieve a high PPV while maintaining good sensitivity. As PPV and NPV depend on prevalence, PPV can be modest in low prevalence even with highly specific assays, whereas NPV remains high. Therefore, interpreting positive results in SPF herds should warrant further investigation, including confirmatory testing and consideration of additional information such as serotype, herd history, and risk.\u003c/p\u003e \u003cp\u003eMultiple APP serotypes can be detected with a single ELISA, provided that the assay incorporates polysaccharide-based surface antigens from the relevant serotypes. However, such assays do not allow differentiation between individual serotypes within the same test. In contrast, microsphere-based multiplex fluorescent immunoassays enable the simultaneous and distinct measurement of antibodies to multiple serotypes or serogroups. This approach reduces sample volume, assay time, labor, and inter-assay variability. When serotype-specific data are required, MFIA provides a more efficient and cost-effective alternative to running multiple individual ELISAs. Additionally, MFIAs have demonstrated the potential to detect lower amounts of analyte present in samples than conventional ELISAs (Baker et al., \u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2012\u003c/span\u003e; Powell et al., \u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e2013\u003c/span\u003e; Wagner and Freer, \u003cspan citationid=\"CR52\" class=\"CitationRef\"\u003e2009\u003c/span\u003e). Notably, sera that tested negative by the ApxIV ELISA were found to be positive using an MFIA based on the same antigen (Gim\u0026eacute;nez-Lirola et al., \u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e2014\u003c/span\u003e). In the present study, the MFIA showed higher sensitivity than both ELISAs, although its performance was largely comparable to that of the 1\u0026ndash;12 ELISA. However, it should be emphasized that our evaluation considered MFIA performance at the assay level rather than for individual serogroups, which can obscure potential variation in performance among antigen-specific components.\u003c/p\u003e"},{"header":"5. Conclusion","content":"\u003cp\u003eThis study demonstrates that the MFIA and the 1\u0026ndash;12 ELISA provide high sensitivity and specificity for APP in Norwegian swine herds, whereas the ApxIV ELISA shows comparatively lower sensitivity and may be less suitable as a stand-alone screening tool in low-prevalence settings. While adjusting the ApxIV cut-off may improve its sensitivity, further evaluation is needed before recommending changes to routine practice. Overall, our findings support the value of multiplex immunoassays for APP surveillance and highlight the importance of validating diagnostic tests within the populations where they will be applied to ensure accurate and effective disease monitoring.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003eAnimal Ethics statement: Approval from the Norwegian Food Safety Authority was not required, as the study did not involve animal experimentation as defined by the Regulation on the Use of Animals in Experiments (FOR 2015 06 18 761). Blood samples from ‘Små i Ro’ were collected during exsanguination in connection with routine slaughter at a commercial abattoir. Samples from the genetic nucleus herd consisted of pre-existing, stored serum samples collected as part of routine health screening of breeding animals.\u003c/p\u003e \u003ch2\u003eDeclaration of Competing Interest\u003c/h2\u003e \u003cp\u003eThe authors collaborate with Biovet on a separate diagnostic test development project; however, the company had no role in any aspect of the evaluation reported in this paper. All assessments were conducted independently by the authors. The authors declare no additional competing financial or personal interests that could appear to influence the work reported in this paper.\u003c/p\u003e \u003cp\u003eAuthor contributions\u003c/p\u003e\n\u003cp\u003eKatharine R. Dean: Methodology, Software, Data curation, Formal analysis, Visualization, Writing-Original Draft\u003c/p\u003e\n\u003cp\u003eKari Lybeck: Conceptualization, Resources,\u0026nbsp;Investigation, Writing-Original Draft\u003c/p\u003e\n\u003cp\u003eIngunn Anita Samdal: Conceptualization,\u0026nbsp;Investigation,\u0026nbsp;Writing - Review \u0026amp; Editing\u003c/p\u003e\n\u003cp\u003eAnniken Jerre Borge:\u0026nbsp;Formal analysis, Writing - Review \u0026amp; Editing\u003c/p\u003e\n\u003cp\u003eElisabeth Skatvedt Jordal: Conceptualization, Resources,\u0026nbsp;Writing - Review \u0026amp; Editing\u003c/p\u003e\n\u003cp\u003eCarl Andreas Gr\u0026oslash;ntvedt: Conceptualization, Resources, Writing \u0026ndash; Review \u0026amp; Editing\u003c/p\u003e\n\u003cp\u003eSondre Stokke Naadland: Conceptualization, Writing \u0026ndash; Review \u0026amp; Editing\u003c/p\u003e\n\u003cp\u003eKristin Udjus:\u0026nbsp;Investigation,\u0026nbsp;Writing - Review \u0026amp; Editing\u003c/p\u003e\n\u003cp\u003eIrene Haugen:\u0026nbsp;Investigation,\u0026nbsp;Writing - Review \u0026amp; Editing\u003c/p\u003e\n\u003cp\u003eSiv Klevar: Conceptualization, Resources, Project administration, Investigation, Writing - Review \u0026amp; Editing\u003c/p\u003e\u003ch2\u003eAcknowledgements\u003c/h2\u003e \u003cp\u003eThis study received financial support from the Research Council of Norway and the Research Funding for Agriculture and the Food Industry (FFL/JA) through project number 326686 PreparePig.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eArnal Bernal JL, Gottschalk M, Lacotoure S, Sanz Tejero C, Chac\u0026oacute;n P\u0026eacute;rez G, Mart\u0026iacute;n-Jurado D, Fern\u0026aacute;ndez Ros AB (2024) Serotype diversity of \u003cem\u003eActinobacillus pleuropneumoniae\u003c/em\u003e detected by real-time PCR in clinical and subclinical samples from spanish pig farms during 2017\u0026ndash;2022. Vet Res 55:165. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1186/s13567-024-01419-2\u003c/span\u003e\u003cspan address=\"10.1186/s13567-024-01419-2\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBaker HN, Murphy R, Lopez E, Garcia C (2012) Conversion of a capture ELISA to a Luminex xMAP assay using a multiplex antibody screening method. J Vis Exp 4084. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3791/4084\u003c/span\u003e\u003cspan address=\"10.3791/4084\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBerger SS, Lauritsen KT, Boas U, Lind P, Andresen LO (2017) Simultaneous detection of antibodies to five \u003cem\u003eActinobacillus pleuropneumoniae\u003c/em\u003e serovars using bead-based multiplex analysis. J Vet Diagn Invest 29:797\u0026ndash;804. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1177/1040638717719481\u003c/span\u003e\u003cspan address=\"10.1177/1040638717719481\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBlackall PJ, Klaasen HLBM, Van Den Bosch H, Kuhnert P, Frey J (2002) Proposal of a new serovar of \u003cem\u003eActinobacillus pleuropneumoniae\u003c/em\u003e: serovar 15. Vet Microbiol 84:47\u0026ndash;52. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/S0378-1135(01)00428-X\u003c/span\u003e\u003cspan address=\"10.1016/S0378-1135(01)00428-X\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBoekema BKHL, Kamp EM, Smits MA, Smith HE, Stockhofe-Zurwieden N (2004) Both ApxI and ApxII of \u003cem\u003eActinobacillus pleuropneumoniae\u003c/em\u003e serotype 1 are necessary for full virulence. Vet Microbiol 100:17\u0026ndash;23. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.vetmic.2003.09.024\u003c/span\u003e\u003cspan address=\"10.1016/j.vetmic.2003.09.024\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBoss\u0026eacute; JT, Janson H, Sheehan BJ, Beddek AJ, Rycroft AN, Kroll JS, Langford PR (2002) \u003cem\u003eActinobacillus pleuropneumoniae\u003c/em\u003e: pathobiology and pathogenesis of infection. Microbes Infect 4:225\u0026ndash;235. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/s1286-4579(01)01534-9\u003c/span\u003e\u003cspan address=\"10.1016/s1286-4579(01)01534-9\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBroes A, Martineau G-P, Gottschalk M (2007) Dealing with unexpected \u003cem\u003eActinobacillus pleuropneumoniae\u003c/em\u003e serological results. J Swine Health Prod 15:264\u0026ndash;269. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.54846/jshap/524\u003c/span\u003e\u003cspan address=\"10.54846/jshap/524\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCaya I, Bertrand M, Broes A (2014) A multiplexed fluorometric immunoassay (MFIA) for the detection of antibodies to \u003cem\u003eActinobacillus pleuropneumoniae\u003c/em\u003e 1-9-11, 2, 3-6-8-15, 4\u0026ndash;7, 5, 10 and 12, in: Proceedings of the American Association of Veterinary Laboratory Diagnosticians Annual Conference. p. 188\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChristopher-Hennings J, Araujo KPC, Souza CJH, Fang Y, Lawson S, Nelson EA, Clement T, Dunn M, Lunney JK (2013) Opportunities for bead-based multiplex assays in veterinary diagnostic laboratories. J Vet Diagn Invest 25:671\u0026ndash;691. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1177/1040638713507256\u003c/span\u003e\u003cspan address=\"10.1177/1040638713507256\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCohen LM, Gr\u0026oslash;ntvedt CA, Klem TB, Gulliksen SM, Ranheim B, Nielsen JP, Valheim M, Kielland C (2020) A descriptive study of acute outbreaks of respiratory disease in Norwegian fattening pig herds. Acta Vet Scand 62:35. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1186/s13028-020-00529-z\u003c/span\u003e\u003cspan address=\"10.1186/s13028-020-00529-z\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCosta G, Oliveira S, Torrison J, Dee S (2011) Evaluation of \u003cem\u003eActinobacillus pleuropneumoniae\u003c/em\u003e diagnostic tests using samples derived from experimentally infected pigs. Vet Microbiol 148:246\u0026ndash;251. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.vetmic.2010.08.023\u003c/span\u003e\u003cspan address=\"10.1016/j.vetmic.2010.08.023\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDenwood MJ (2016) runjags: an R package providing interface utilities, model templates, parallel computing methods and additional distributions for MCMC models in JAGS. J Stat Softw 71:1\u0026ndash;25. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.18637/jss.v071.i09\u003c/span\u003e\u003cspan address=\"10.18637/jss.v071.i09\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDreyfus A, Schaller A, Nivollet S, Segers RP, a. M, Kobisch M, Mieli L, Soerensen V, H\u0026uuml;ssy D, Miserez R, Zimmermann W, Inderbitzin F, Frey J (2004) Use of recombinant ApxIV in serodiagnosis of \u003cem\u003eActinobacillus pleuropneumoniae\u003c/em\u003e infections, development and prevalidation of the ApxIV ELISA. Vet Microbiol 99:227\u0026ndash;238. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.vetmic.2004.01.004\u003c/span\u003e\u003cspan address=\"10.1016/j.vetmic.2004.01.004\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDubreuil JD, Jacques M, Mittal KR, Gottschalk M (2000) \u003cem\u003eActinobacillus pleuropneumoniae\u003c/em\u003e surface polysaccharides: their role in diagnosis and immunogenicity. Anim Health Res Rev 1:73\u0026ndash;93. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1017/S1466252300000074\u003c/span\u003e\u003cspan address=\"10.1017/S1466252300000074\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eEamens G, Gonsalves J, Whittington A-M, Turner B (2012) Evaluation of serovar-independent ELISA antigens of \u003cem\u003eActinobacillus pleuropneumoniae\u003c/em\u003e in pigs following vaccination or experimental challenge with respiratory pathogens and natural \u003cem\u003eA. pleuropneumoniae\u003c/em\u003e serovar 1 challenge. Aust Vet J 90:490\u0026ndash;498. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1111/j.1751-0813.2012.01008.x\u003c/span\u003e\u003cspan address=\"10.1111/j.1751-0813.2012.01008.x\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eEn\u0026oslash;e C, Andersen S, S\u0026oslash;rensen V, Willeberg P (2001) Estimation of sensitivity, specificity and predictive values of two serologic tests for the detection of antibodies against \u003cem\u003eActinobacillus pleuropneumoniae\u003c/em\u003e serotype 2 in the absence of a reference test (gold standard). Prev Vet Med 51:227\u0026ndash;243. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/S0167-5877(01)00226-4\u003c/span\u003e\u003cspan address=\"10.1016/S0167-5877(01)00226-4\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFuresz SE, Mallard BA, Boss\u0026eacute; JT, Rosendal S, Wilkie BN, MacInnes JI (1997) Antibody- and cell-mediated immune responses of \u003cem\u003eActinobacillus pleuropneumoniae\u003c/em\u003e-infected and bacterin-vaccinated pigs. Infect Immun 65:358\u0026ndash;365. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1128/iai.65.2.358-365.1997\u003c/span\u003e\u003cspan address=\"10.1128/iai.65.2.358-365.1997\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGardner IA, Stryhn H, Lind P, Collins MT (2000) Conditional dependence between tests affects the diagnosis and surveillance of animal diseases. Prev Vet Med 45:107\u0026ndash;122. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/S0167-5877(00)00119-7\u003c/span\u003e\u003cspan address=\"10.1016/S0167-5877(00)00119-7\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGim\u0026eacute;nez-Lirola LG, Jiang Y-H, Sun D, Hoang H, Yoon K-J, Halbur PG, Opriessnig T (2014) Simultaneous detection of antibodies against apx toxins ApxI, ApxII, ApxIII, and ApxIV in pigs with known and unknown \u003cem\u003eActinobacillus pleuropneumoniae\u003c/em\u003e exposure using a multiplexing liquid array platform. Clin Vaccine Immunol 21:85\u0026ndash;95. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1128/CVI.00451-13\u003c/span\u003e\u003cspan address=\"10.1128/CVI.00451-13\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGonz\u0026aacute;lez W, Gim\u0026eacute;nez-Lirola LG, Holmes A, Lizano S, Goodell C, Poonsuk K, Sitthicharoenchai P, Sun Y, Zimmerman J (2017) Detection of \u003cem\u003eActinobacillus pleuropneumoniae\u003c/em\u003e ApxIV toxin antibody in serum and oral fluid specimens from pigs inoculated under experimental conditions. J Vet Res 61:163\u0026ndash;171. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1515/jvetres-2017-0021\u003c/span\u003e\u003cspan address=\"10.1515/jvetres-2017-0021\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGottschalk M (2015) The challenge of detecting herds sub-clinically infected with \u003cem\u003eActinobacillus pleuropneumoniae\u003c/em\u003e. Vet J 206:30\u0026ndash;38. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.tvjl.2015.06.016\u003c/span\u003e\u003cspan address=\"10.1016/j.tvjl.2015.06.016\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGottschalk M (2012) Actinobacillus. In: Zimmerman JJ, Karriker LA, Ramirez A, Schwartz KJ, Stevenson GW (eds) Diseases of Swine. Wiley-Blackwell, Ames, IA, pp 653\u0026ndash;669\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGottschalk M, Altman E, Charland N, De Lasalle F, Dubreuil JD (1994) Evaluation of a saline boiled extract, capsular polysaccharides and long-chain lipopolysaccharides of \u003cem\u003eActinobacillus pleuropneumoniae\u003c/em\u003e serotype 1 as antigens for the serodiagnosis of swine pleuropneumonia. Vet Microbiol 42:91\u0026ndash;104. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/0378-1135(94)90009-4\u003c/span\u003e\u003cspan address=\"10.1016/0378-1135(94)90009-4\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGottschalk M, Altman E, Lacouture S, De Lasalle F, Dubreuil JD (1997) Serodiagnosis of swine pleuropneumonia due to \u003cem\u003eActinobacillus pleuropneumoniae\u003c/em\u003e serotypes 7 and 4 using long-chain lipopolysaccharides. Can J Vet Res 61:62\u0026ndash;65\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGreiner M, Gardner IA (2000) Epidemiologic issues in the validation of veterinary diagnostic tests. Prev Vet Med 45:3\u0026ndash;22. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/S0167-5877(00)00114-8\u003c/span\u003e\u003cspan address=\"10.1016/S0167-5877(00)00114-8\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGr\u0026oslash;ntvedt CA, Jordal ES, Valheim M, Urdahl AM, Ellingsen-Dalskau K (2023) Svin. Dyrehelserapporten 2022. Veterin\u0026aelig;rinstituttet\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGulliksen SM, Baustad B, Framstad T, J\u0026oslash;rgensen A, Skoms\u0026oslash;y A, Kjelvik O, Gjestvang M, Gr\u0026oslash;ntvedt CA, Lium B (2021) Successful eradication of \u003cem\u003eMycoplasma hyopneumoniae\u003c/em\u003e from the Norwegian pig population \u0026ndash; 10 years later. Porc Health Manag 7:37. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1186/s40813-021-00216-z\u003c/span\u003e\u003cspan address=\"10.1186/s40813-021-00216-z\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHui SL, Walter SD (1980) Estimating the Error Rates of Diagnostic Tests. Biometrics 36:167\u0026ndash;171. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.2307/2530508\u003c/span\u003e\u003cspan address=\"10.2307/2530508\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eIDEXX (2011) APP-Apx IV Ab test validation data report. IDEXX Laboratories Inc\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eIDvet (2018) Internal validation report ID Screen\u0026reg; APP Screening Indirect. Innovative Diagnostics\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eInnovative, Diagnostics (2026) Personal communication via e-mail\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKlausen J, Ekeroth L, Gr\u0026oslash;ndahl-Hansen J, Andresen LO (2007) An indirect enzyme-linked immunosorbent assay for detection of antibodies to \u003cem\u003eActinobacillus pleuropneumoniae\u003c/em\u003e serovar 7 in pig serum. J Vet Diagn Invest 19:244\u0026ndash;249. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1177/104063870701900303\u003c/span\u003e\u003cspan address=\"10.1177/104063870701900303\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKostoulas P, Nielsen SS, Branscum AJ, Johnson WO, Dendukuri N, Dhand NK, Toft N, Gardner IA (2017) STARD-BLCM: Standards for the Reporting of Diagnostic accuracy studies that use Bayesian Latent Class Models. Prev Vet Med 138:37\u0026ndash;47. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.prevetmed.2017.01.006\u003c/span\u003e\u003cspan address=\"10.1016/j.prevetmed.2017.01.006\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKwan W-F, Li Y, Boss\u0026eacute; JT, Chiou M-T, Chiu H-J, Langford PR, Mortensen P, Lin C-N (2025) Serovars and antimicrobial resistance profiles of \u003cem\u003eActinobacillus pleuropneumoniae\u003c/em\u003e isolates from clinical-case pigs in Taiwan. BMC Vet Res 21:502. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1186/s12917-025-04878-7\u003c/span\u003e\u003cspan address=\"10.1186/s12917-025-04878-7\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMidtveit I (2023) Helse p\u0026aring; sm\u0026aring;gris i Rogaland \u0026ndash; ein statusrapport fr\u0026aring; prosjektet Sm\u0026aring;-i-Ro. Svin\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMidtveit I (2022) Status for sm\u0026aring;griskvaliteten i Rogaland. Bondevennen\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNorsvin SA (2026) Friskere gris med SPF - Norsvin [WWW Document]. URL \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://norsvin.no/friskere-gris-med-spf/\u003c/span\u003e\u003cspan address=\"https://norsvin.no/friskere-gris-med-spf/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (accessed 2.26.26)\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eOlsen A, Nielsen HV, Alban L, Houe H, Jensen TB, Denwood M (2022) Determination of an optimal ELISA cut-off for the diagnosis of \u003cem\u003eToxoplasma gondii\u003c/em\u003e infection in pigs using Bayesian latent class modelling of data from multiple diagnostic tests. Prev Vet Med 201:105606. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.prevetmed.2022.105606\u003c/span\u003e\u003cspan address=\"10.1016/j.prevetmed.2022.105606\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eOpriessnig T, Hemann M, Johnson JK, Heinen S, Gim\u0026eacute;nez-Lirola LG, O\u0026rsquo;Neill KC, Hoang H, Yoon K-J, Gottschalk M, Halbur PG (2013) Evaluation of diagnostic assays for the serological detection of \u003cem\u003eActinobacillus pleuropneumoniae\u003c/em\u003e on samples of known or unknown exposure. J Vet Diagn Invest 25:61\u0026ndash;71. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1177/1040638712469607\u003c/span\u003e\u003cspan address=\"10.1177/1040638712469607\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePaulina P, Dawid T (2025) Serotyping and antimicrobial resistance of \u003cem\u003eActinobacillus pleuropneumoniae\u003c/em\u003e isolates from fattening pigs in Poland from 2019 to 2024. BMC Vet Res 21:40. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1186/s12917-025-04504-6\u003c/span\u003e\u003cspan address=\"10.1186/s12917-025-04504-6\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePlummer M (2025) rjags: Bayesian graphical models using MCMC\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePowell RLR, Ouellette I, Lindsay RW, Parks CL, King CR, McDermott AB, Morrow G (2013) A multiplex microsphere-based immunoassay increases the sensitivity of SIV-specific antibody detection in serum samples and mucosal specimens collected from rhesus macaques infected with SIVmac239. BioResearch Open Access 2. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1089/biores.2013.0009\u003c/span\u003e\u003cspan address=\"10.1089/biores.2013.0009\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e. biores.2013.0009\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCore Team R (2025) R: a language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSassu EL, Boss\u0026eacute; JT, Tobias TJ, Gottschalk M, Langford PR, Hennig-Pauka I (2018) Update on \u003cem\u003eActinobacillus pleuropneumoniae\u003c/em\u003e\u0026mdash;knowledge, gaps and challenges. Transbound Emerg Dis 65:72\u0026ndash;90. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1111/tbed.12739\u003c/span\u003e\u003cspan address=\"10.1111/tbed.12739\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSchaller A, Djordjevic SP, Eamens GJ, Forbes WA, Kuhn R, Kuhnert P, Gottschalk M, Nicolet J, Frey J (2001) Identification and detection of \u003cem\u003eActinobacillus pleuropneumoniae\u003c/em\u003e by PCR based on the gene apxIVA. Vet Microbiol 79:47\u0026ndash;62. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/s0378-1135(00)00345-x\u003c/span\u003e\u003cspan address=\"10.1016/s0378-1135(00)00345-x\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSj\u0026ouml;lund M, Zoric M, Persson M, Karlsson G, Wallgren P (2011) Disease patterns and immune responses in the offspring to sows with high or low antibody levels to \u003cem\u003eActinobacillus pleuropneumoniae\u003c/em\u003e serotype 2. Res Vet Sci 91:25\u0026ndash;31. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.rvsc.2010.07.025\u003c/span\u003e\u003cspan address=\"10.1016/j.rvsc.2010.07.025\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSoto Perezchica MM, Guerrero Barrera AL, Avelar Gonzalez FJ, Quezada Tristan T, Marin M, O (2023) \u003cem\u003eActinobacillus pleuropneumoniae\u003c/em\u003e, surface proteins and virulence: a review. Front Vet Sci 10. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.3389/fvets.2023.1276712\u003c/span\u003e\u003cspan address=\"10.3389/fvets.2023.1276712\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eStringer OW, Boss\u0026eacute; JT, Lacouture S, Gottschalk M, Fodor L, Angen \u0026Oslash;, Velazquez E, Penny P, Lei L, Langford PR, Li Y (2021) Proposal of \u003cem\u003eActinobacillus pleuropneumoniae\u003c/em\u003e serovar 19, and reformulation of previous multiplex PCRs for capsule-specific typing of all known serovars. Vet Microbiol 255:109021. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.vetmic.2021.109021\u003c/span\u003e\u003cspan address=\"10.1016/j.vetmic.2021.109021\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eStygar AH, Niemi JK, Oliviero C, Laurila T, Heinonen M (2016) Economic value of mitigating \u003cem\u003eActinobacillus pleuropneumoniae\u003c/em\u003e infections in pig fattening herds. Agric Syst 144:113\u0026ndash;121. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.agsy.2016.02.005\u003c/span\u003e\u003cspan address=\"10.1016/j.agsy.2016.02.005\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTegetmeyer HE, Jones SCP, Langford PR, Baltes N (2008) ISApl\u003cem\u003e1\u003c/em\u003e, a novel insertion element of \u003cem\u003eActinobacillus pleuropneumoniae\u003c/em\u003e, prevents ApxIV-based serological detection of serotype 7 strain AP76. Vet Microbiol 128:342\u0026ndash;353. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.vetmic.2007.10.025\u003c/span\u003e\u003cspan address=\"10.1016/j.vetmic.2007.10.025\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eToft N, J\u0026oslash;rgensen E, H\u0026oslash;jsgaard S (2005) Diagnosing diagnostic tests: evaluating the assumptions underlying the estimation of sensitivity and specificity in the absence of a gold standard. Prev Vet Med 68:19\u0026ndash;33. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.prevetmed.2005.01.006\u003c/span\u003e\u003cspan address=\"10.1016/j.prevetmed.2005.01.006\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWagner B, Freer H (2009) Development of a bead-based multiplex assay for simultaneous quantification of cytokines in horses. Vet Immunol Immunopathol 127:242\u0026ndash;248. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1016/j.vetimm.2008.10.313\u003c/span\u003e\u003cspan address=\"10.1016/j.vetimm.2008.10.313\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWickham H (2016) ggplot2: Elegant Graphics for Data Analysis. Springer-, New York\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWickham H, Averick M, Bryan J, Chang W, McGowan LD, Fran\u0026ccedil;ois R, Grolemund G, Hayes A, Henry L, Hester J, Kuhn M, Pedersen TL, Miller E, Bache SM, M\u0026uuml;ller K, Ooms J, Robinson D, Seidel DP, Spinu V, Takahashi K, Vaughan D, Wilke C, Woo K, Yutani H (2019) Welcome to the tidyverse. J Open Source Softw 4:1686. \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.21105/joss.01686\u003c/span\u003e\u003cspan address=\"10.21105/joss.01686\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWickham H, Fran\u0026ccedil;ois R, Henry L, M\u0026uuml;ller K, Vaughan D (2023) dplyr: A Grammar of Data Manipulation\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWorld Organisation for Animal Health (2021) Principles and methods of validation of diagnostic assays for infectious diseases [WWW Document]. Terrestrial Manual. URL \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.woah.org/fileadmin/Home/fr/Health_standards/tahm/1.01.06_VALIDATION.pdf\u003c/span\u003e\u003cspan address=\"https://www.woah.org/fileadmin/Home/fr/Health_standards/tahm/1.01.06_VALIDATION.pdf\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[{"identity":"9f0f14bc-2195-4005-bc4e-81bc5273fffd","identifier":"10.13039/501100005416","name":"Norges Forskningsråd","awardNumber":"326686","order_by":0}],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":true,"highlight":"","institution":"Norwegian Veterinary Institute","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"APP, BLCM, test evaluation, diagnostic accuracy, sensitivity, specificity","lastPublishedDoi":"10.21203/rs.3.rs-9608302/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-9608302/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003e \u003cem\u003eActinobacillus pleuropneumoniae\u003c/em\u003e (APP) remains a major cause of respiratory disease in swine and accurate serological tools are essential for surveillance, particularly in sub-clinical herds and low prevalence settings. We evaluated the diagnostic performance of a microsphere-based multiplex fluorescent immunoassay (MFIA) and two commercial ELISAs (1\u0026ndash;12 ELISA and ApxIV ELISA) using a Bayesian latent class model and field samples from Norwegian swine herds. Although both ELISAs and the MFIA showed high specificity, the ApxIV ELISA demonstrated lower sensitivity (89.4%) than the other assays. These findings align with previously reported variability in the sensitivity of the ApxIV-based ELISA. The MFIA showed high sensitivity, demonstrating the potential of multiplex immunoassays for APP serology. We also explored optimization of the ApxIV ELISA cut-off, which indicated that lowering the threshold could improve the sensitivity of the test without compromising the specificity for our samples. Our estimates can be influenced by several factors, including the serotype diversity in the population and assumptions related to latent class modelling. Despite these constraints, our findings support the use of MFIA and the ELISAs as accurate tools for APP surveillance, while highlighting that the ApxIV ELISA may be less suitable as a stand-alone screening test. These results underscore the importance of validating diagnostic assays in populations where they will be applied to ensure that they are fit-for-purpose.\u003c/p\u003e","manuscriptTitle":"Evaluation of a multiplex assay and two commercial ELISAs for the serological detection of Actinobacillus pleuropneumoniae using Bayesian latent class modelling","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-05-11 02:20:42","doi":"10.21203/rs.3.rs-9608302/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"2cbd26d7-ae69-46fd-b247-e9223657ad39","owner":[],"postedDate":"May 11th, 2026","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2026-05-11T02:20:43+00:00","versionOfRecord":[],"versionCreatedAt":"2026-05-11 02:20:42","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-9608302","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-9608302","identity":"rs-9608302","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}