An update on FIV and FeLV test performance using a Bayesian statistical approach

BACKGROUND: Screening tests for feline retroviruses are thought to have high sensitivity and specificity, although previous studies that evaluated these tests have limitations. Novel statistical approaches have been developed that allow the estimation of sensitivity and specificity in situations where the true state of the disease in individual animals cannot be assured. OBJECTIVE: The purpose of this study was to evaluate the sensitivity and specificity of a variety of retrovirus tests, including some screening tests, in a population of cats potentially infected with either feline leukemia virus (FeLV) and/or feline immunodeficiency virus (FIV) by using a Bayesian statistical approach. METHODS: Four hundred and ninety blood samples from cats being evaluated for FIV infection were tested by 2 rapid immunomigration tests (Witness single [WS], Witness combi [WC]) and a plate-based ELISA (Petcheck) for FIV antibody, and by a newly designed real-time polymerase chain reaction (PCR) assay for FIV provirus. Four hundred and ninety-five blood samples from cats being evaluated for FeLV infection were tested by 2 rapid immunomigration tests (WS, WC) and a plate-based ELISA (Petcheck) for FeLV antigen, and by a FeLV virus isolation technique. Results were then analyzed by using a Bayesian statistical method. RESULTS: For FIV tests, median sensitivity estimates were 0.98 for WS, 0.97 for WC, 0.98 for ELISA, and 0.92 for PCR. Median specificity estimates were 0.96 for WS, 0.96 for WC, 0.93 for ELISA, and 0.99 for PCR. For FeLV tests, median sensitivity estimates were 0.97 for WS, 0.97 for WC, 0.98 for ELISA, and 0.91 for virus isolation. Median specificity estimates were 0.96 for WS, 0.96 for WC, 0.98 for ELISA, and 0.99 for virus isolation. CONCLUSIONS: The use of Bayesian statistical methods overcomes a variety of methodologic problems associated with diagnostic test evaluations, including the lack of a definitive reference test. The sensitivity and the specificity of all 6 evaluated screening tests was high: however, specificity estimates were slightly lower than those reported by most recent studies.     

Preliminary evaluation of diagnostic tests for avian influenza using the Markov Chain Monte Carlo (MCMC) Method in an emergency surveillance

In June 2005, an outbreak of avian influenza (AI) caused by a low pathogenic H5N2 virus was identified in Japan. A serological surveillance was conducted because the infected chickens did not show any clinical signs. The Markov Chain Monte Carlo Method was used to evaluate the performances of serological HI and AGP tests because there was not enough time when the surveillance was initiated to conduct a test evaluation. The sensitivity of the AGP test (0.67) was lower than that of the HI test (0.99), while the specificities were high for both tests (0.96 for AGP and 0.90 for HI). Based on the low sensitivity of the AGP test, the HI test was used for primary screening in later stages of the epidemic.     

Development of enzyme-linked immunosorbent assay with nucleoprotein as antigen for detection of antibodies to avian influenza virus

During the avian influenza outbreak of 2003-04 in Southeast Asia, two avian influenza viruses (AIV), one of H5N1 subtype and the other H9N2 subtype, were isolated and identified from local farms. The nudeoprotein (NP) gene of the H5N1 AI isolate was cloned, and the segment encoding amino acid 47-384, which covers its major antigenic domains, was subcloned and expressed in E. coli. Subsequently, the NP (47-384) expression product was purified and used as the diagnostic antigen to develop a NP-based type-specific indirect enzyme-linked immunosorbent assay (ELISA) for detecting antibodies to AI from chicken sera. The ELISA is shown to be specific for AIV and does not cross-react with chicken sera that has antibodies to other avian viruses. The NP(47-384)-ELISA was compared with a hemagglutination inhibition test and a commercial AIV ELISA kit in evaluating 150 sera samples from experimentally AIV-infected or vaccinated specific-pathogen-free (SPF) chickens. Our NP(47-384)-ELISA was more sensitive than the two tests and showed an 82% agreement ratio with the HI test and an 80.67% agreement ratio with the commercial kit. The NP(47-384)-ELISA and the commercial AIV ELISA were used to evaluate 448 field sera samples from diseased chickens or vaccinated chickens during the 2003-04 AI outbreak in China. The two ELISA tests had a 95% agreement ratio. We conclude that the NP(47-384)-ELISA developed in our laboratory was specific and sensitive and it has great application potential in China's long-term prevention and control of AI.     

Using latent class analysis to estimate the test characteristics of the γ-interferon test, the single intradermal comparative tuberculin test and a multiplex immunoassay under Irish conditions

Considerable effort has been devoted to improving the existing diagnostic tests for bovine tuberculosis (single intradermal comparative tuberculin test [SICTT] and γ-interferon assay [γ-IFN]) and to develop new tests. Previously, the diagnostic characteristics (sensitivity, specificity) have been estimated in populations with defined infection status. However, these approaches can be problematic as there may be few herds in Ireland where freedom from infection is guaranteed. We used latent class models to estimate the diagnostic characteristics of existing (SICTT and γ-IFN) and new (multiplex immunoassay [Enferplex-TB]) diagnostic tests under Irish field conditions where true disease status was unknown. The study population consisted of herds recruited in areas with no known TB problems (2197 animals) and herds experiencing a confirmed TB breakdown (2740 animals). A Bayesian model was developed, allowing for dependence between SICTT and γ-IFN, while assuming independence from the Enferplex-TB test. Different test interpretations were used for the analysis: SICTT (standard and severe interpretation), γ-IFN (a single interpretation), and a range of interpretations for the Enferplex-TB (level-1 [high sensitivity interpretation] to level-5 [high specificity interpretation]). The sensitivity and specificity (95% posterior credibility intervals; 95% PCI) of SICTT[standard] relative to Enferplex-TB[level-1] and γ-IFN were 52.9-60.8% and 99.2-99.8%, respectively. Equivalent estimates for γ-IFN relative to Enferplex-TB[level-1] and SICTT were 63.1-70.1% and 86.8-89.4%, respectively. Sensitivity of Enferplex-TB[level-1] (95% PCI: 64.8-71.9%) was superior to the SICTT[standard], and specificity of the Enferplex-TB[level-5] was superior to γ-IFN (95% PCI: 99.6-100.0%). These results provide robust measures of sensitivity and specificity under field conditions in Ireland and suggest that the Enferplex-TB test has the potential to improve on current diagnostics for TB infection in cattle. The extent of that potential will be assessed in further studies.     

Unexpected high responses to tuberculin skin-test in farmed red deer: implications for tuberculosis control

Tuberculosis (TB) in deer is a serious zoonotic disease of worldwide distribution. Detection of infected animals is usually performed using single or comparative skin-testing (SST/CST), although false responses due to sensitization to other mycobacteria may occur, hampering diagnostic specificity. We describe the evolution of the responses to the SST, CST and to an in-house serological assay in a red deer farm subjected to regular TB testing in southern Spain in an attempt to understand the dynamics of possible non-specific reactions occurring under field conditions. We performed 2288 skin-tests and ELISAs in nine sampling periods between May 2009 and January 2011. In May 2010, a strong increase in skin fold thickness in response to avian purified protein derivative (PPD) (mean=4.0mm, 95% CI=3.5-4.5) and bovine PPD (mean=1.8mm, 95% CI=1.6-2.0) was observed in yearling deer hinds (n=150), compared to values recorded for the same individuals in November 2009 (avian PPD: mean=0.7 mm, 95% CI=0.6-0.8 and bovine PPD: mean=0.7 mm, 95% CI=0.6-0.7) and in January 2011 (avian PPD: mean=2.2mm, 95% CI=1.9-2.4 and bovine PPD: mean=1.1mm, 95% CI=1.0-1.2). Using SST, 54 animals (36%) of the yearlings tested in May 2010 would have been classified as positive reactors, while none of them was positive in the CST. The five animals with highest skin fold increases to mycobacterial antigens were culled and subjected to post-mortem analysis, which confirmed the absence of Mycobacterium tuberculosis complex (MTBC) infection but demonstrated the presence of environmental mycobacteria and closely related bacteria in four out of the five analyzed animals. Our results demonstrated how non-specific responses to mycobacterial antigens can adversely affect the specificity of TB diagnosis based on the SST. Thus, once TB infection has been ruled out using confirmatory techniques, application of comparative diagnostic tests is highly advisable to maximize test specificity and avoid the slaughter of false positive reactors.     

Antimicrobial resistance of Escherichia coli isolated from chickens

Faecal Escherichia coli isolated from healthy farm chickens, from farm chickens with avian influenza, and from chickens with diarrhoea were more resistant to antimicrobial agents (94-100%) than those isolated from healthy domestic chickens (20%). Transfer of drug resistance was readily achieved from strains isolated from both healthy and sick farm chickens, and from diarrhoeic chickens; it was more difficult to demonstrate in strains from domestic chickens. Resistant E. coli showing serotypes suspected to be enteropathogenic for man, i.e 0126:K71(B16), 044:K74 (L) and 0119:K69(B14), were isolated from faecal samples of healthy and sick farm chickens, but not from healthy domestic birds.     

The sensitivity and specificity of single manual examination tests for detection of corpora lutea in the ovaries of non-cycling cows

OBJECTIVE: To evaluate the sensitivity and specificity of single manual examination of ovarian surface structures and of ovarian size as tests for the detection of corpora lutea in non-cycling cows DESIGN: Non cycling cows were dichotomously classified for the presence of palpable corpora lutea and for ovarian size. The performance characteristics of the palpable corpus luteum test and the ovarian size test were estimated using Bayesian analysis. Previously published information on prevalence and test performance was combined with observed data using the Gibbs Sampler to derive posterior distributions for prevalence and test parameters. RESULTS: Prior distributions for the prevalence of corpora lutea in cows not detected on heat before mating within seasonal herds were centred on 25%, and for the corpus luteum test sensitivity were centred on 70%. No prior assumptions for any other test parameter were made. From a total of 650 cows, 144 were found to have at least one corpus luteum, and 156 were found to have two small ovaries. The posterior estimate obtained for prevalence was 0.30 (95% Cl 0.22 to 0.43), for corpus luteum test sensitivity was 0.71 (95% Cl 0.49 to 0.93), and for corpus luteum test specificity was 0.98 (95% Cl 0.49 to 1.00). For the ovarian size test, the posterior estimate for size test sensitivity was 0.98 (95% Cl 0.95 to 1.00) and for size test specificity was 0.34 (95% CI 0.28 to 0.43). Sensitivity analysis indicated that corpus luteum test sensitivity ranged from 0.60 to 0.70 in most herds. CONCLUSIONS: Errors associated with pre-mating heat detection are likely to result in between 15 to 30% of cycling cows not detected in oestrous before mating, resulting in the inclusion of cycling cows within the population of non-cycling cows. This mixed population of cows is then subjected to manual examination by veterinarians in order to assign cows to treatment groups. The corpus luteum test has modest sensitivity and high specificity and the size test has high sensitivity and low specificity. Therefore the use of a single examination of the ovaries of cows not detected on heat before the mating period, in order to classify them as anovulatory anoestrous cows with or without a corpus luteum, is not sufficient to accurately classify them and thus to recommend treatment.     

Evaluation of the diagnostic accuracy of the modified agglutination test (MAT) and an indirect ELISA for the detection of serum antibodies against Toxoplasma gondii in sheep through Bayesian approaches

The diagnostic accuracies of the modified agglutination test (MAT) and indirect ELISA test for the detection of serum antibodies against Toxoplasma gondii in sheep were evaluated through Bayesian approaches on two populations of sheep created from three different groups of animals (T. gondii-aborted ewes, colostrums-deprived newborn lambs, and ewe-lambs and adult ewes with unknown T. gondii infection status). Tests showed a high degree of agreement (kappa statistic = 0.93; 95% confidence interval = 0.87, 0.98) and a significant specificity (Sp) correlation (gamma(Sp) = 0.26; 95% credibility interval = 0.017, 0.61). When prior information was used for all unknown parameters the posterior medians for the sensitivity (Se) and Sp of the MAT and ELISA were, respectively, 92.6% (95% credibility interval = 85.2, 96.9), 95.5% (89.9, 98.7), 90.5% (83.4, 95.6), and 97.8% (94.2, 99.5). These estimates remained similar when uninformative priors were included. The Se estimates of the MAT and ELISA were higher than those obtained on pigs in other study using the same approach (Se = 80.6% and Sp = 89.5% for the MAT, and Se = 71.5% and Sp = 85.5% for the ELISA [Georgiadis, M.P., Wesley, O.J., Gardner, I.A., Singh, R., 2003. Correlation-adjusted estimation of sensitivity and specificity of two diagnostic tests. Appl. Stat. 52, 63-78]. This finding supported the believe that test performances may vary when applied on different animal species. Thus, if these tests are planned to be used on animal species other than sheep or pigs, their diagnostic accuracy should be re-assessed to prevent biased inferences from their results.     

Seroprevalence and risk factors of swine influenza in Spain

Swine influenza is caused by type A influenza virus. Pigs can be infected by both avian and human influenza viruses; therefore, the influenza virus infection in pigs is considered an important public health concern. The aims of present study were to asses the seroprevalence of swine influenza subtypes in Spain and explore the risk factors associated with the spread of those infections. Serum samples from 2151 pigs of 98 randomly selected farms were analyzed by an indirect ELISA for detection of antibodies against nucleoprotein A of influenza viruses and by the hemagglutination inhibition (HI) using H1N1, H1N2 and H3N2 swine influenza viruses (SIV) as antigens. Data gathered in questionnaires filled for each farm were used to explore risk factors associated with swine influenza. For that purpose, data were analyzed using the generalized estimating equations method and, in parallel by means of a logistic regression. By ELISA, 92 farms (93.9%; CI(95%): 89.1-98.7%) had at least one positive animal and, in total, 1340/2151 animals (62.3%; CI(95%): 60.2-64.3%) were seropositive. A total of 1622 animals (75.4%; CI(95%): 73.6-77.2%) were positive in at least one of the HI tests. Of the 98 farms, 91 (92.9%; CI(95%): 87.7-98.1%) had H1N1 seropositive animals; 63 (64.3%; CI(95%): 54.6-73.9%) had H1N2 seropositive pigs and 91 (92.9%; CI(95%): 87.7-98.1%) were positive to H3N2. Mixed infections were detected in 88 farms (89.8; CI(95%): 83.7-95.9%). Three risk factors were associated with seroprevalences of SIV: increased replacement rates in pregnancy units and, for fatteners, existence of open partitions between pens and uncontrolled entrance to the farm.     

An enzyme-linked immunosorbent assay for the detection of antibody against avian influenza virus

An enzyme-linked immunosorbent assay (ELISA) was developed for detecting antibody to type A avian influenza (AI) virus. The sensitivity and group specificity of the AI-ELISA were compared with those of the agar-gel-precipitin test (AGPT) and the hemagglutination-inhibition (HI) test under conditions of both controlled and field exposure. During the course of temporal experimental infection (0-76 days) of specific-pathogen-free (SPF) chickens with AI subtype Hav9N2, the AI-ELISA was able to detect specific AI antibody as early as 8 days postinoculation (PI), and it measured rising levels of antibody through 35 days PI, at which time the chickens were re-exposed to AI virus. Conversely, AGP tests were negative through 35 days PI, and HI tests began to detect low levels of AI antibody only at 21 days PI. Following a secondary infection at 35 days PI with the same AI subtype, all tests measured rising levels of AI-specific antibody (35-76 days PI). However, the AGP test was positive at only the 7- and 14-day samplings postsecondary immunization. Under field conditions, the AI-ELISA was able to detect serum AI antibody in flocks from which highly pathogenic AI was isolated, but the AGP tests of these sera were negative.     

Development of a specific latex agglutination test based on a recombinant hemagglutinin protein to detect antibodies to H5 avian influenza viruses

A latex agglutination test (LAT) was developed for rapid detection of antibodies to H5 avian influenza viruses (AIVs). The hemagglutinin protein of H5 AIV was covalently linked to carboxylated latex by ethyl-dimethyl-amino-propyl carbodiimide to prepare the sensitized latex beads. The LAT was evaluated with the hemagglutination inhibition (HI) assay as the reference test. The H5-LAT showed a sensitivity of 87.0% and specificity of 88.9% in detecting 126 serum samples from experimentally infected chickens and a sensitivity of 82.5% and specificity of 86% in detecting 587 field chicken serum samples from mostly vaccinated chickens. The agreement ratio between H5-LAT and HI was found to be 87.3% and 83.1% for the two groups of samples, respectively. Difficulty with background agglutination in stored chicken sera was overcome by serum pretreatment with either dried chicken liver powder or dilution buffer containing detergent Tween-20. The H5-LAT has advantages over a previously reported whole-virus LAT in terms of biosafety in preparation, chemical stability, and higher specificity. It is a rapid and simple test suitable for field monitoring of antibodies to H5-type AIV.     

Detection of antibodies to the nonstructural protein (NS1) of avian influenza viruses allows distinction between vaccinated and infected chickens

To differentiate avian influenza virus (AIV)-infected chickens vs. chickens immunized with inactivated avian influenza virus, an enzyme-linked immunosorbent assay (ELISA) was developed using a recombinant nonstructural protein (NS1) as the diagnostic antigen, which was cloned from an AIV H9N2 subtype strain isolated during the avian influenza outbreak of 2003-04 and expressed in Escherichia coli. Antibodies to the AIV NS1 protein was only detected in the sera of chickens experimentally infected with AIV but not in the sera of chickens immunized with inactivated vaccine. This ELISA is useful for serological diagnosis to distinguish chickens infected with influenza viruses from those immunized with inactivated vaccine.     

Canine visceral leishmaniosis: a comparative analysis of the EIE-leishmaniose-visceral-canina-Bio-Manguinhos and the IFI-leishmaniose-visceral-canina-Bio-Manguinhos kits

This study evaluated the performance of the EIE-leishmaniose-visceral-canina-Bio-Manguinhos (EIE-LVC) kit and to compare it with that of the IFI-leishmaniose-visceral-canina-Bio-Manguinhos (IFI-LVC) kit. Four groups of dogs were studied: group 1 (G1), dogs with clinical signs indicative of CVL and testing positive for the parasite (n = 25); group 2 (G2), dogs with only a presumed diagnosis of CVL (n = 62); group 3 (G3), dogs that had never lived in an area where CVL is endemic and never received a blood transfusion (n = 16); group 4 (G4), dogs carrying other parasites: such as babesiosis (n = 4), ehrlichiosis (n = 6) and demodicosis (n = 1). G1 and G3 were used for the calculation of sensitivity and specificity, respectively. The EIE-LVC showed a sensitivity of 72% (IC 95%: 50.4-87.1%) and a specificity of 87.5% (IC 95%: 60.4-97.8%). The value of the kappa index was 0.975 (CI 95%: 0.926-1.024), which represents an excellent fit. For IFI-LVC, the sensitivity was 68.0% (CI 95%: 46.4-84.3%) and the specificity 87.5% (CI 95%: 60.4-97.8%). When the tests were conducted in parallel, sensitivity was 92.0% (CI 95%: 72.5-98.6%) and specificity 75.0% (CI 95%: 47.4-91.7%). However, when conducted consecutively, the tests showed a sensitivity of 48.0% (CI 95%: 28.3-68.2%) and a specificity of 100.0% (CI 95%: 75.9-99.4%). The analysis of clinically suspected dogs using IFI-LVC and EIE-LVC kits in parallel, revealed that 26/62 animals were positive. Cross-reaction was observed in a dog with demodicosis. These results lead to the following conclusions: (1) the performance of the EIE-LVC kit is not statistically different from the IFI-LVC and (2) the kits must be used in parallel if higher sensitivity is required, reducing the number of false-negative results.     

Application of real-time RT-PCR utilising lux (light upon extension) fluorogenic primer for the rapid detection of avian influenza viruses

A real-time RT-PCR assay utilising light upon extension fluorogenic primer (LUX RT-PCR) was developed for the rapid and efficient detection of avian influenza viruses (AIV). The assay detected each of the AIV isolates tested (16/16) and gave negative results with heterologous pathogens (17/17). The detection limit of the assay proved to be 10(-0.5) EID50/0.2 ml and 10(1.5) EID50/0.2 ml in allantoic fluid of virus-infected embryonated chicken eggs and in spiked chicken faeces samples, respectively. Based on its specificity, sensitivity and relative simplicity, the LUX RT-PCR assay provides a novel, rapid and cost-effective diagnostic tool for avian influenza surveillance and monitoring programs.     

Establishment of an antibody avidity test to differentiate vaccinated cattle from those naturally infected with Mycoplasma bovis

Mycoplasma bovis is a major pathogen of bovine respiratory disease (BRD) in China and a live attenuated vaccine has recently been developed. This study aimed to establish an IgG avidity test to differentiate between naturally infected and vaccinated animals. An indirect ELISA (iELISA) was first established in the laboratory to detect antibodies specific to M. bovis using whole cell proteins as coating antigens and serum samples from experimentally infected cattle. The specificity and sensitivity of the iELISA was confirmed using a commercial ELISA kit as a reference standard. Both tests showed substantial agreement as indicated by a κ value of 0.78 (95% confidence interval, CI, 0.62, 0.93), and an overall 92.0% (80/87) agreement between the two tests. Based on the laboratory iELISA, a sodium thiocyanate (NaSCN) competitive iELISA was then developed for the detection of IgG avidity, expressed as relative avidity index (AI).Two-hundred and one experimentally immunised and naturally infected animals were used. These comprised 36 immunised calves, 38 negative control calves, 37 naturally infected calves, 87 calves of unknown status, and an additional three immunised calves that were used for a time trial. By testing true positive and negative antisera from either naturally infected or immunised calves, the AI cut-off value was defined as 70.4%. The diagnostic accuracy of the in-house NaSCN competitive iELISA was determined using serum samples collected from the experimental animals. The IgG avidity test demonstrated 96.0% sensitivity (95% CI 80.5%, 99.3%) and 95.8% specificity (95% CI 79.8%, 99.3%), and was successfully established as a valuable first test for differentiating vaccinated animals from those infected with M. bovis. This test may be a useful tool for clarifying the magnitude of M. bovis infection and in assessing the efficacy of vaccination in exposed animal populations.     

Sero-prevalence of avian influenza among broiler-breeder flocks in Jordan

Thirty blood samples were collected randomly from each of the 38 breeder-broiler farms in Jordan. Serum samples were examined using indirect ELISA for specific antibodies to avian influenza virus. The overall true flock-level sero-prevalence of avian influenza was 71% (95% CI: 55,83). Positive flocks had 2-30 sero-positive chickens and half of flocks had >20 sero-positive birds. The number of sero-positive flocks varied in the studied localities with more sero-positives in farms located within the migratory route of migratory wild fowl. The examined broiler-breeder flocks had no clinical signs, or noticeable decrease in egg production; mortalities were within the normal range (0.1-1%). The number of positive sera/flock correlated with flock size. There were a no significant (Pearsons r=0.21, p=0.21) correlation between positive flocks and age. A non-pathogenic AI virus infects broiler-breeder farms in Jordan. Wild local and migrating birds might promote the further spread of this virus in Jordan and other countries.     

PRE- AND POST-TREATMENT ULTRASONOGRAPHY IN HYPOTHYROID DOGS

Primary hypothyroidism is a frequent endocrine disorder in the adult dog. However, false-positive diagnoses are common because of the relatively low accuracy of most commonly used biochemical tests. The purpose of this study was to describe the ultrasonographic features of the thyroid gland in hypothyroid dogs, to calculate the diagnostic sensitivity of gray-scale ultrasound using a combination of clinical symptoms and biochemical thyroid tests as gold standard, and to investigate the evolution of the ultrasonographic features after treatment of hypothyroidism. Eighteen dogs were studied prospectively. All dogs underwent an ultrasound examination at first presentation and 13 underwent one or two additional ultrasound examinations over time. At first presentation, a sensitivity of 76.5% (95% CI [50.0-93.0% 1) for decreased echogenicity, 64.7% (95% CI [38.385.8% 1) for inhomogeneity, 70.6% (95% CI 144.0-89.7%]) for irregular capsule delineation, 64.7% (95% CI [38.3-85.8%]) for abnormal lobe shape and 47.1% (95% CI 123.0-72.2%]) for decreased relative thyroid volume was obtained. Combining these five parameters together resulted in an overall sensitivity of 94.1% (95% CI [71.3-99.9%]) for gray-scale ultrasound in the detection of acquired hypothyroidism at first presentation. A continuous decrease of thyroid volume was seen over time after treatment, while the other investigated parameters did not change significantly during the follow-up period. None of the thyroid glands were considered normal at the last presentation. Grayscale ultrasound is a sensitive and quick test for the diagnosis of primary hypothyroidism in dogs.