Clinical signs and pathology
During the course of the experiment, all pigs infected with PCV2 in combination with either PPV or PRRSV developed clinical disease signs with a varying degree of severity. No systemic disease signs were observed in the control group with the exception of a temporary lameness in one pig and paleness in two pigs. Three pigs in the PCV2/PRRSV infected group and one pig from the PCV2/PPV infected group were euthanized for humane reasons at 25 and 26 days post infection. Two pigs died directly after blood sampling, supposedly not related to the experimental infection; one from the control group (at 18 days post infection) and another pig from the PCV2/PRRSV group (at 12 days post infection).
Pigs in both the PCV2/PRRSV and the PCV2/PPV infected groups showed signs of wasting with a significant difference in weekly body weight gain compared to the control group. This was consistently seen starting at one week post infection (p < 0.001 and p < 0.05, respectively). Body weight gain was significantly lower in the PCV2/PRRSV infected group compared to the PCV2/PPV infected group in the first week post infection. In the second week post infection, weight gain was similar in both infected groups. Data are shown in Figure .
Figure 1 Body weight gain (A), Body temperatures (B) and clinical scores (C) in experimentally infected groups. Course of the body weight gain (Figure 1A), body temperature (Figure 1B) and clinical scores (Figure 1C) were recorded in experimental groups of three (more ...)
Mean rectal temperature in pigs in the PCV2/PRRSV infected group increased to febrile temperatures, i.e. rectal temperatures ≥ 40.0°C from two days post infection on for seven days and again elevated mean body temperatures were seen at 14 days post infection and for a period of five days between 18 and 23 days post infection (Figure ). In the PCV2/PPV infected group febrile body temperatures were observed incidentally at seven days post infection and at 13 days post infection.
Clinical signs started to appear between day six and seven in all pigs from the PCV2/PRRSV infected group and also in a number of pigs from the PCV2/PPV infected group. In the PCV2/PRRSV group the mean clinical score, based on the occurrence and severity of clinical symptoms, reached 2.5 (moderate to severe disease) of a maximum score of 3. In this group severe disease signs started to develop from seven days post infection on (Figure ). In the PCV2/PPV infected group the manifestation of disease symptoms occurred slightly later and the mean clinical score was generally lower in this group. Clinical signs as severe depression were observed in more than 80% of all pigs from the PCV2/PRRSV group compared to 10% in the PCV2/PPV group. Whereas respiratory distress was recorded in all infected pigs, 75% from the PRRSV co-infected group showed signs of pneumonia and only about 20% of the PPV co-infected group (Figure ). A palpable increase of the size of the inguinal lymph nodes was found in all PCV2/PRRSV infected pigs from 12 days post infection on and in three pigs from the PCV2/PPV infected group from 12, 19 or 21 days post infection on.
Figure 2 Clinical symptoms in experimentally infected groups. The percentage of animals in three weeks old, colostrum-deprived piglets showing diverse signs of disease. Observations were made twice daily during a time frame of 26 days post inoculation with either (more ...)
At necropsy, typically for PMWS, inguinal lymph nodes and also other lymph nodes were enlarged in all infected pigs and the mean weights of the inguinal lymph nodes in both infected groups were higher than in the control group (PCV2/PRRSV vs. control, p < 0.03; PCV2/PPV versus control, p < 0.08). In the PCV2/PRRSV infected group two of nine pigs had a macroscopically identifiable pneumonia, although a moderate to severe interstitial pneumonia was found in eight of nine pigs based on histology. In the PCV2/PPV infected group macroscopic changes were restricted to increased size of the inguinal lymph node and kidney, liver or lung changes in a few pigs. Histologically, most striking was a slight to moderate hepatitis in seven of eight pigs. In five of 17 PCV2 infected pigs lymph node depletion was observed, in others a hyperplasia was more prominent.
After termination of the experiment, tissue samples were tested for the presence of PRRSV and PCV2 nucleic acid detection by PCR. All infected animals showed strong positive results for PCV2 in lymph nodes, lung and spleen. Although in control animals no PCV2 nucleic acid was detected in pharyngeal swabs throughout the study, low levels of PCV2 nucleic acid were found in single or several tissue specimens of six control animals. PRRSV nucleic acid was detected in lungs of all PCV2/PRRSV infected animals but not of the other groups. No PCR testing has been performed for PPV.
Acute phase proteins
Results of acute phase proteins levels are summarized in Figure . At day five p.i. levels of acute phase proteins did not differ significantly compared to levels in sera collected prior to inoculation, except for pig major protein (PigMAP, p = 0.004) and albumin (p = 0.01) in the control group and PigMAP in the PCV2/PPV infected group (p = 0.035). More significant differences were observed at day 19 p.i. compared to levels at day zero for three acute phase proteins; haptoglobin (Hp) in the PCV2/PRRSV group (p < 0.001), PigMAP in PCV2/PPV (p = 0.023) and PCV2/PRRSV infected animals (p = 0.021), and albumin in PCV2/PPV (p = 0.002), PCV2/PRRSV infected animals (p = 0.003), as well as in the PBS treated control group (p = < 0.001).
Figure 3 Acute phase protein levels in experimentally infected groups. Mean levels of C-reactive protein (CRP), serum amyloid A(SAA), haptoglobin (Hp), pig major protein (PigMAP), α-Lipoprotein (ApoA1), and albumin were determined in three weeks old, colostrum-deprived (more ...)
Some statistically significant differences were observed comparing different experimental groups. At day zero, PigMAP levels in animals from the PCV2/PRRSV group (p = 0.013) and α-Lipoprotein (ApoA1) levels in the PCV2/PPV group (p < 0.001) were significantly different compared to the control group. At day five p.i., only ApoA1 (control group versus PCV2/PPV group, p = 0.004), and at day 19 only albumin (control group versus PCV2/PRRSV group, p = 0.032) were discriminative.
Serum protein profiles were obtained on day zero as well as on day five post infection, before clinical symptoms became apparent. A third set of protein profiles was obtained at day 19 post infection, when all animals from both experimental infections showed clear signs of disease. Using the Protein chip Data Manager software a total of 586 protein peaks were identified and subjected to further statistical analysis. It should be noted that these 586 protein peaks may represent a lower number of proteins, as a certain overlap may be present among the results obtained with the three types of arrays that have been used.
Protein profiles of serum obtained from all animals prior to infection (day 0) were tested for differences. No statistical significant differences could be observed between animal groups, indicating that in this respect, the three groups of animals were very similar at the time of infection.
Overview of comparisons
We performed multiple analyses, comparing the three experimental groups: animals inoculated with (i) PCV2 and PPV (PCV2/PPV); (ii) PCV2 and PRRSV (PCV2/PRRSV); and (iii) PBS (controls), at three time points. An overview of these analyses is given in Table . The comparisons can be categorized as follows: (1) infected animals (PCV2/PPV together with PCV2/PRRSV) versus non-infected control animals, (2) PCV2/PPV infected animals versus control animals or PCV2/PRRSV infected animals versus control animals, (3) PCV2/PPV, PCV2/PRRSV, and control animals as three distinct groups to explore the discriminatory power of serum protein profiles to distinguish the different infection models.
Overview of the statistical analyses (A-H) with the accompanying data sets that were used
For each comparison the number of significantly different protein peaks in their profiles was determined as well as the classification accuracy. Additionally we tested whether reducing the number of protein peaks in a profile affected the accuracy of classification. Next to whole protein profiles consisting of 586 protein peaks, we also tested profiles with 500, 200, 100, 50, 20, and 10 protein components. For this, proteins were ranked by significance and those with highest significance were selected.
Number of significant different protein peaks
For each comparison the number of significantly different protein peaks is shown in Table . Based on a maximum p-value of 0.01, the number of differentially expressed protein components ranged from 15 (PCV2/PPV versus control) to 59 (PCV2/PRRSV versus control). For a number of comparisons, molecular masses of the most significant differently expressed protein components are summarized in Table .
Overview of the number of significant differentially expressed protein components in different analyses
Most significantly differentially expressed protein components as characterized by mass:charge (m/z) value
As expected, based on the False Discovery Rate (FDR), which may be more appropriate than p-values as it accounts for multiple testing, the amount of differentially expressed protein components was reduced. In a number of analyses, none of the protein components showed a significantly different expression based on an FDR < 0.05. In general, more protein components were differentially expressed comparing PCV2/PRRSV infected animals versus non-infected control animals as opposed to PCV2/PPV infected versus control animals as shown in Table . Combining data of day five and day 19 post infection (p.i.) increased the number of significantly differentially expressed protein components.
Infected animals versus non-infected animals
In this analysis, PCV2/PPV along with PCV2/PRRSV infected animals were marked as infected, whereas control animals at day five and day 19 are regarded as non-infected (Table ). Classification results are summarized in Table .
The number of animals correctly classified as either infected or non-infected
At day 19 post infection, both PCV2/PPV and PCV2/PRRSV infected animals displayed evident signs of illness with 14 of 15 (Sensitivity (Se) = 93.3%; specificity (Sp) = 57.1%) infected animals having significantly different serum protein profiles compared to non-infected animals. Moreover at day five, before any disease symptoms were apparent, 14 of 16 (Se = 87.5%; Sp = 66.7%) infected animals could be recognized based on their serum protein profiles.
Interestingly, preselecting protein peaks slightly increased the classification accuracy; using the 50 (day five post infection) or 100 (day 19 post infection) most significant differently expressed proteins resulted in the highest number of correctly classified animals. When the number of proteins was further limited to ten, a decrease in correctly classified animals was observed at day five (Se = 62.5%;Sp = 44.4%). However at day 19 post infection, profiles based on the ten most significant proteins could still identify 93% (= Se) of the infected animals, although only one of the seven non-infected animals was correctly classified (Sp = 14.3%) (Table ).
Two-group classification: PCV2/PPV versus control or PCV2/PRRSV versus control
Serum protein profiles of PCV2/PPV infected animals were compared with profiles of non-infected control animals on day five and day 19 using either data from a single time point or combining data from both days. Similarly, profiles of PCV2/PRRSV infected animals were compared with control animals. Results are summarized in Table .
Number of animals correctly classified according to infection status, evaluating two distinct classes
With regard to the classification of infected versus non-infected animals at day five post infection, results were poor for the PCV2/PPV group as only one of seven (Se = 14.3%; Sp = 55.6%) infected animals could be distinguished from control animals by profiles consisting of 586 protein components. The classification performance of serum protein profiles was much better for PCV2/PRRSV infected animals (Se = 88%; Sp = 77.8%) at day five post infection.
As expected, overall results were better at day 19 post infection. Six of eight (Se = 75%; Sp = 28.6%) PCV2/PPV and five of seven (Se = 71.4%; Sp = 71.4%) PCV2/PRRSV infected animals could be distinguished from non-infected control animals by serum protein profiles using 586 protein components. This reflects the clinical signs, which were very similar for animals in both groups at 19 days post infection.
Combining the data of both days slightly increased the classification accuracy as compared to profiles from either day five or day 19 post infection. This is especially true for the control groups as shown in Table (Sp ranges from 66.7% - 100%).
Again, using a preselection of the most significantly different protein peaks generally led to comparable or even better classification accuracy, similarly as described for the classification of infected versus non-infected animals.
To assess the power of serum protein profiles as a diagnostic marker for specific infections, we explored the classification performance of serum protein profiles to distinguish between the three distinct animal groups (i) non-infected control animals, (ii) PCV2/PRRSV and (iii) PCV2/PPV infected animals) in one analysis, i.e. three-group classification. As expected, three-group classification as shown in Table gave similar results compared to the two-group classification, but with lower classification accuracy.
Number of animals correctly classified according to infection status, evaluating three distinct classes
Table shows the contingency tables for day five and 19 for true disease status and classification based on serum protein profiling using 50 most significant protein peaks. At day five, almost all (8/9) PCV2/PRRSV infected animals were classified correctly. A single PCV2/PRRSV infected animal was misclassified as PCV2/PPV infected. On the other hand, only two of seven PCV2/PPV animals were identified correctly. One was improperly classified as PCV2/PRRSV infected, while four animals could not be discriminated from non-infected animals. However at 19 days post infection, five of eight PCV2/PPV infected animals were accurately classified, while only three of seven PCV2/PRRSV infected animals could be identified based on their serum protein profiles. The other four animals were misclassified as PCV2/PPV infected. Control animals, when misclassified, were labeled as PCV2/PPV infected animals, but never as PCV2/PRRSV infected.
Contingency table showing the classification results according to infection status, evaluating three distinct classes
In conclusion, PCV2/PRRSV infected animals could well be distinguished from control animals as early as day five, while PCV2/PPV infections were best distinguished from non-infected control animals at day 19. Classification results showed a very high significance at day five (P < 0.001) and were near significance at day 19 (P = 0.095). It can be concluded that based on SELDI-TOF protein profiles, at day five post infection PCV2/PRRSV infected animals are easier to distinguish compared to PCV2/PPV infected animals.