Results from T. gondii and T. spiralis negative field samples, negative control serum of the experimental infection and secondary antibody binding alone (Table ) showed that the height of NSB responses is foremost dependent of the presence of serum. Observing that negative sera due to NSB can reach the same or higher responses as for example positive control serum of 250,000 MFI on T. gondii and 500,000 MFI on T. spiralis beads (data not presented), it is concluded that correction for NSB is necessary to prevent false positive results.
Non specific binding responses of T.gondii and T. spiralis
negative serum sets and conjugate alone
The results of T. gondii and T. spiralis negative field sera illustrate that the response of the reference beads could not directly be used as a measure for non-specific binding. The log linear relations between responses of reference beads and T. gondii or T. spiralis beads are depicted in Figures and , and the relations were expressed as:
Figure 1 Estimation of T. gondii non-specific binding correction factor. Log transformed responses of T. gondii negative swine field sera on non-coupled beads (x-axis) versus T. gondii coupled beads (y-axis). Linear regression line (log y = 0.404* log x + 2.818), (more ...)
Figure 2 Estimation of T. spiralis non-specific binding correction factor. Log transformed responses of T. spiralis negative swine field sera on non-coupled beads (x-axis) versus T. spiralis coupled beads (y-axis). Linear regression line (log y = 0.646*log x + (more ...)
The slopes of the two regression lines indicate that T. gondii beads are subject to less NSB as compared to T. spiralis beads. This finding may be explained by, for example, variable concentration of antigens on the bead, differences in antigen molecule structures and orientation on the bead surface and/or the affinity between non-specific antibodies and the unoccupied bead surface or coupled antigens. A variety in antigen composition between biologically produced batches can therefore be of importance in relation to NSB. To test whether the correction factor to calculate bead correlated NSB is stable between batches of antigen, further evaluation is necessary. Subtraction of uncorrected responses of reference beads, according to formula 1, would lead to an underestimation or overestimation of NSB for responses below and above 53,480 and 183,110 for T. gondii and T. spiralis, respectively.
Results of ROC calculations, presented in Table , showed that the AUC, a measure of agreement between specific responses and the experimental infection status of the animals, of all indirect tests ranged between 0.837 and 0.930 for T. gondii and 0.855 and 0.879 for T. spiralis. These values indicate that there is a good relation between the responses of all indirect tests and the infection status of the animals.
Infection status based specifications of bead-based assay and ELISA tests calculated by ROC analysis
A perfect test is a test in which the responses correspond 100%, i.e. an AUC of 1.000, with the values of the test to which it is compared. The imperfect AUC values (< 1.000) found in our study (Table ) can partly be explained by a late immunological development of antibodies, which is associated to the time course of parasite antigen expression and the immune response of infected animals. Evidence from earlier studies showed that muscle larvae, depending on the infection dose, can be found in pork by digestion or trichinoscopy as early as 17 days p.i. [13
]. Other studies showed that T. spiralis
ES could be measured within the developing muscle larvae and its cuticular surface as early as 14 days p.i. [14
] and in the surrounding tissue around 15 days p.i. [15
]. Consequently, the response time, i.e. the time of development of antibodies against the antigen used in the indirect T. spiralis
assays, is affected by this late production. Porcine IgG antibodies against ES are developed approximately 3 to 4 weeks after infection with 5,000 muscle larvae [16
]. Porcine IgG antibodies against T. gondii
tachyzoites are produced much earlier in time and can be detectable after one to two weeks of infection [18
]. In our study, the sera used for ROC calculations originated from animals which were collected on a weekly basis [8
]. Samples drawn 5 days after T. gondii
inoculation and 5, 12 and 19 days after T. spiralis
inoculation would produce a false negative result when compared to the experimental infection status, resulting in lower AUC values. Calculations of ROC curves without these sera resulted in notably higher AUC values of 0.995, 0.999, 0.998 and 0.999 for BBA-TOX, E3-TOX, BBA-TRI and E-TRI, respectively (Figures and ).
Inter-test agreement between assays calculated by McNemar's and Cohen's Kappa analysis
Figure 3 ROC curves of T. gondii assays calculated with a limited serum set from experimental infection. ROC analysis of a T. gondii bead-based assay and three ELISAs using responses from a set of serum samples, consisting of sera from experimentally infected (more ...)
Figure 4 ROC curves of T. spiralis assays calculated with a limited serum set from experimental infection. ROC analysis of a T. spiralis bead-based assay and ELISA using responses from a set of serum samples, consisting of sera from experimentally infected pigs (more ...)
According to data of test performances and inter test agreement, presented in Table , the T. gondii and T. spiralis bead assays agree excellently with their respective highest scoring tests, i.e. with the E-TOX3 and E-TRI, respectively. Marginal homogeneity by McNemars test, a test which determines the equality between positive and negative test proportions of one test compared to the other, indicated that there is a balance between BBA-TOX and E-TOX3.
The potential use of the bead based test is prevention of T. gondii
and/or T. spiralis
infections in pork to enter the human food chain. Nonetheless, direct parasitological testing, like T. spiralis
artificial digestion, are more reliable methods to test for present infection in meat. Existing T. gondii
direct tests are either laborious, e.g. due to the need of pathogen extraction in PCR [19
], or are undesirable because of utilization of experimental animals, like in cat and mice bioassays [20
], to determine the infectious status of meat. In the case of T. gondii
infections, serological testing is the next best option to perform on large scale. Unfortunately, due to the time window between infection and development of specific antibody responses, serological tests are less reliable for detection on individual scale; however, they can be used for monitoring purposes on herd level [21
To prevent human T. gondii
and T. spiralis
infections through consumption of infected pork by serological monitoring of pig herds, a high sensitivity of 99% [22
], and an approximation of AP to TP (Table ) is desired. None of the assays used for this paper met this requirement. However, when ROC calculations were restricted to serum samples in which antibody responses were to be expected, as was described above, the sensitivity was 97% and 99% for BBA-TOX and BBA-TRI, respectively (data not presented). Subsequent calculations for true and apparent prevalence resulted in an overall T. gondii
TP of 52.7% and AP's of 51.5% and 53.3% for BBA and ELISA, respectively. The overall TP of T. spiralis
was 44.4% while the APs were 44.1% and 44.3% for BBA and ELISA, respectively (data not presented). These data would indicate that the combined bead-based assay is applicable for serological monitoring purposes.
Comparison of true and apparent prevalence's of pig sera between indirect assays
Although we compared our new T. gondii
and T. spiralis
bead test with a limited selection of available ELISAs, the test specifications and agreement between tests examined in this study indicate that the combined bead test equals or is superior to other tests. However, all calculations have been based upon tests using serum samples of experimentally infected pigs, which were exposed to high doses of parasites. Conventionally raised animals are likely to be infected by lower doses of parasites, and sero-conversion may be detected later in time [16
]. Therefore, to determine the applicability of this bead-based test for the use of indirect detection of infection, it is advisable to further evaluate the test by the use of serum samples of naturally infected pigs.