Introduction

The use of sampling methods alternative to blood is especially relevant in pigs due to the high stress and pain that blood collection causes in this species. One of the samples that is currently being increasingly used due to its non-invasive nature is saliva or oral fluid. The term “saliva” usually refers to the secretion of the salivary glands, while “oral fluid” refers to saliva plus other components such as crevicular fluid, cellular debris and microorganisms; although in literature the two terms are often used interchangeably. In pigs, this kind of sample is especially used for the detection of infectious agents. In addition, this sample can also be used to evaluate different conditions and physiologic aspects, including stress, inflammation, immune response, redox homeostasis, and tissue damage1,2,3.

In this line, profiles integrated by various biomarkers can be used in saliva of pigs. Stress can be assessed with analytes including α-amylase (AA) and butyrylcholinesterase (BChE), whereas the presence of inflammation can be detected with analytes such as C-reactive protein (CRP), haptoglobin (Hp), total proteins (TP), myeloperoxidase (MPO) and ferritin. Also, changes in immune system with biomarkers like adenosine deaminase (ADA) and immunoglobulin G (IgG) can be evaluated. Redox status can be assessed with analytes such as uric acid (UA), the ferric reducing ability (FRA) of saliva or advanced oxidation protein products (AOPP), and tissue damage with the measurement of enzymes such as creatine kinase (CK), aspartate aminotransferase (AST) and lactate dehydrogenase (LDH). Other routine biochemistry analytes can also give valuable health status information, such as the enzymes alkaline phosphatase (ALP) or γ-glutamil transferase (gGT), metabolites such as Triglycerides or Urea, or minerals like Calcium and Phosphorus3.

One of the aspects of interest for the interpretation of the analytes in saliva is their possible correlation with serum. This has been described to be high in cases of analytes in which there is a passive diffusion from blood to salivary gland, such as urea4 whereas in other analytes, such as ADA, there is a lack of correlation between saliva and serum, possibly because of local production in salivary gland5. In this line, studies that can provide information about the possible relations in the analytes between saliva and serum would be of interest to elucidate the possible mechanisms of the presence of analytes in saliva and help in their interpretation.

The aims of this study were: (1) to make a comparative evaluation of the potential influence of several factors, such as age or gender on the biomarkers in saliva and serum of healthy pigs along with a correlation study between the two fluids.; and (2) to compare the response of these biomarkers in saliva and serum in a disease condition, such as Actinobacillus pleuropneumoniae (A. pp.) infection. For this purpose, a comprehensive profile of analytes including AA, BChE, CRP, Hp, TP, MPO, ferritin, ADA, IgG, UA, FRA, AOPP, CK, AST, LDH, ALP, gGT, Trig, urea, Ca and P were measured in saliva and serum at different times during a fattening cycle in healthy pigs and also in pigs with A. pp. infection.

Results

The results obtained in the age, sex and sample type interactions appear in Table 1. Since no significance was found when evaluating the potential interaction between sex of the animals and either sample type or animal age in healthy animals, the variable sex was excluded from the statistical analyses. Therefore, all the data presented includes all animals used in the study, integrating males and females as a group.

Table 1 Global significances of the mixed linear model when evaluating the interaction of the type of sample, sex and the age of the pigs.
Full size table

Differences in various analytes were found when evaluating the interaction between sample type and animal age, as well as between sample type and health status. Detailed pairwise comparisons evaluating those interactions are provided in Tables 2, 3, 4 and 5.

Table 2 Median (interquartile range) concentrations of stress and immune system biomarkers in saliva and serum in the different ages of collection (T0 = 30 days, T1 = 55 days, T2 = 115 days-old) and animals with Actinobacillus pleuropneumoniae infection (Diseased). Different superscript letters (a, b, c) indicate p < 0.05 between ages of recollection, whereas same letters indicate p > 0.05 and the absence of letters denotes p > 0.05 across all time points. Asterisks indicate significant difference between T1 and diseased (*: p < 0.05; **: P < 0.01; ***: p < 0.001; ns: non-significant).
Full size table
Table 3 Median (interquartile range) concentrations of inflammatory biomarkers in saliva and serum in the different ages of collection (T0 = 30 days, T1 = 55 days, T2 = 115 days-old) and animals with Actinobacillus pleuropneumoniae infection (Diseased). Different superscript letters (a, b, c) indicate p < 0.05 between ages of recollection, whereas same letters indicate p > 0.05 and the absence of letters denotes p > 0.05 across all time points. Asterisks indicate significant difference between T1 and diseased (*: p < 0.05; **: P < 0.01; ***: p < 0.001; ns: non-significant).
Full size table
Table 4 Median (interquartile range) concentrations of redox status and muscle damage biomarkers in saliva and serum in the different ages of collection (T0 = 30 days, T1 = 55 days, T2 = 115 days-old) and animals with Actinobacillus pleuropneumoniae infection (Diseased). Different superscript letters (a, b, c) indicate p < 0.05 between ages of recollection, whereas same letters indicate p > 0.05 and the absence of letters denotes p > 0.05 across all time points. Asterisks indicate significant difference between T1 and diseased (*: p < 0.05; **: P < 0.01; ***: p < 0.001; ns: non-significant).
Full size table
Table 5 Median (interquartile range) concentrations of other routine biochemistry analytes in saliva and serum in the different ages of collection (T0 = 30 days, T1 = 55 days, T2 = 115 days-old) and animals with Actinobacillus pleuropneumoniae infection (Diseased). Different superscript letters (a, b, c) indicate p < 0.05 between ages of recollection, whereas same letters indicate p > 0.05 and the absence of letters denotes p > 0.05 across all time points. Asterisks indicate significant difference between T1 and diseased (*: p < 0.05; **: P < 0.01; ***: p < 0.001; ns: non-significant).
Full size table

The effect of age and sample type in healthy animals

When changes in analytes were evaluated during different times of a fattening cycle, differences between the dynamics in saliva versus serum were seen in all analytes, with the exception of FRA, Urea, Triglycerides and Ca that showed a similar evolution in saliva and serum during the different timepoints. Although Sample type*Age interaction was significant for AA and P, a similar behavior was seen in saliva and serum after pairwise Bonferroni correction. Similarly, the interaction was not significant in the case of Hp; however, a different dynamic was seen between saliva and serum after Bonferroni correction.

In saliva, most of the analytes showed higher values at T0 with the exception of UA, which had higher values at T1 compared to T0 and T2, and AA, CRP, Ferritin, FRA, and P that did not show statistically significant differences across time points.

In serum, the highest values at T0 were seen in ADA, AOPP, FRA, LDH, ALP, gGT, Ca, Triglycerides and Urea. On the other hand, the analytes that achieved higher values at T1 were CRP, Ferritin, AST and CK. At T2, higher values were observed in IgG and TP compared to T0 and T1. No significant differences were seen in AA, BChE, Hp, MPO and P serum levels along the different time points. UA provided values in serum under the LLOD of the assay.

Effect of sample type in the values of analytes in healthy and diseased animals.

When biomarker levels were compared between healthy and diseased animals, different dynamics were seen between saliva and serum in several analytes, with the exception of TP, Ferritin, CK, gGT, Urea and P. TP, Ferritin, Urea and P showed a similar change, with higher values in diseased animals, in both sample types. Conversely CK and gGT did not show differences between healthy and diseased pigs neither in serum nor saliva. Although MPO and AST did not show a significant Sample type*Health condition interaction, a different dynamic was seen in those analytes after Bonferroni correction.

Both serum and saliva samples showed higher values in diseased animals than in healthy ones for IgG, CRP and Hp, however, the magnitude of these differences varied between the sample types. IgG and CRP showed a higher response in saliva (6.8 and 5.9-fold, respectively) than in serum (2.1 and 4.0-fold, respectively); whereas in Hp the response was slightly higher in saliva (4.4-fold) than in serum (3.8-fold). On the other hand, different behaviour was seen between saliva and serum in AOPP, FRA and Triglycerides, since concentration of these analytes were higher in saliva of the diseased animals compared to healthy ones (3.5, 2.6 and 1.5-fold, respectively), whereas in serum their concentrations were lower in diseased animals than in healthy (0.5, 0.7 and 0.7-fold, respectively).

In the cases of AA, ADA, MPO, AST, LDH and Ca, changes were seen only in saliva, with values significantly higher in diseased animals (3.3, 3.1, 2.2, 2.8, 7.1 and 1.4-fold, respectively) than in healthy ones. Whereas levels of BChE and ALP in the serum of diseased animals were respectively 0.5- and 0.4-fold, lower than those observed in healthy ones, with no differences found in saliva between both groups of animals.

The levels of UA in serum were all under the LLOD of the assay, so only salivary levels were studied. Salivary UA did not show differences between both groups.

The results of the ROC analyses in saliva and serum are shown in Table 6; Fig. 1. The results showed that 17/21 analytes in saliva, and 13/21 in serum, had the ability to discriminate between healthy and diseased animals. In saliva, both stress biomarkers (AA and BChE) significantly discriminated between groups, but whereas AA was not useful in serum, BChE in serum showed a better performance than in saliva. The best performance in saliva was seen for the immune system biomarkers ADA and IgG, both showing AUC > 0.9; in serum, IgG also showed AUC > 0.9 but ADA was not significant. In inflammatory biomarkers, all analytes showed AUC with significant power to discriminate between healthy and diseased pigs, with AUC values in both saliva and serum higher than 0.7, with the exception of MPO in serum, which showed non-significant values. The redox status biomarkers AOPP and FRA had significant AUC in both saliva and serum. In contrast, salivary UA was not able to discriminate diseased from healthy animals. Regarding tissue damage biomarkers, AST and LDH in saliva showed a significant AUC, with values higher than 0.7. The analytes Urea, Triglycerides, Ca and P showed a significant AUC in both saliva and serum, being Urea the best one with AUC > 0.9 in both sample types. Serum ALP also showed a significant discriminatory power with AUC > 0.9, but it was not significant in saliva.

Table 6 Receiver operating characteristic (ROC) analyses for discriminating diseased animals (A. pp infected, n = 20) from healthy animals (T1, n = 30). The area under the curve (AUC) is expressed followed by standard error within parenthesis. CI: confidence interval.
Full size table
Fig. 1
figure 1

Receiver operating characteristic (ROC) curves obtained with the different analytes in saliva (A) and serum (B) for discriminating diseased animals (A. pp infected, n = 20) from healthy animals (T1, n = 30).

Full size image

Spearman correlation between serum and saliva

Table 7 shows the Spearman correlation coefficients obtained between saliva and serum values. The correlation saliva – serum was mild for IgG and Urea, and weak for ADA, CRP, Hp and Ca. No significant or negligible correlations were obtained for the rest of the analytes.

Table 7 Spearman correlation coefficients (r) and significance level between saliva and serum (n = 110).
Full size table

Discussion

To study the changes of the analytes in saliva and serum with age in the pigs of this report, we selected three times of sampling in order to cover the main different phases of the fattening period: after weaning (30 days); an intermediate sample within the nursery phase (50–60 days); and an intermediate sample within the fattening period (115 days). These times have been previously used in other studies about changes in analytes during fattening5.

Sex of the animals was a factor that did not significantly influence the results of this report. However, this contrasts with previous studies that reported higher values of some analytes in females, such as TP, AOPP, UA, FRAS, AST, CK, ALP, Ca, P, gGT and Urea6. The reason for these differences is not clear. Nevertheless, some of these analytes showed a trend (not significant) to be at higher levels in saliva of females than in males at nursery such as TP, AOPP, UA, FRAS (data not shown). Further studies should be performed to clarify how sex can influence the interpretation of salivary biomarkers.

Age-related changes in analytes across the different times in saliva revealed that most analytes showed higher values at initial sampling times. This could be influenced by the changes occurring in piglets in weaning, a phase marked by significant stress and immune challenges. Mixing with other pigs, intestinal alterations caused by the transition from maternal milk to solid feed, as well as the loss of passive immunity from maternal antibodies are components of these challenges6,7.​ These stressors and immune demands can profoundly influence salivary biomarkers, according to the results of another previous research5.

In serum, there were less analytes that showed an increase at the initial time points compared to saliva and many biomarkers that did not significantly change along the productive cycle, such as serum amylase, Hp, TP, MPO and P. BChE showed a low level in serum at the initial time point as it has been previously described in other studies8,9.

With the exception of UA, CK and gGT, the rest of the analytes showed changes in the diseased group in either saliva and/or serum. The change in AA and BChE could be related with the stress induced by the respiratory distress, whereas the changes in CRP, Hp, IgG and ADA could indicated the presence of inflammation as well as immune system activation that has been described in respiratory tract associated with A. pp.10. In addition, the observed change in the redox biomarkers would indicate oxidative stress, since in this disease the activation of immune system cells such as macrophages leads to release of oxygen metabolites enhancing lung tissue damage11; although the mechanism was probably different in saliva and serum, since they were increased in saliva and decreased in serum. The increase in LDH in diseased animals could indicate muscle damage possibly due to inflammation or hypoxia; although it was observed only in saliva that was therefore more sensitive than serum for this purpose. Finally, although further studies should be performed to elucidate the mechanisms of these changes, it could be postulated that the higher Urea values in the diseased pigs could reflect dehydration, and the higher Ca and P values could be due to a reduced glomerular filtration rate.

In a disease condition consisting of an infection with A. pp, the comparison of saliva and serum in disease conditions within this study reveals four different possible patterns:

  1. a.

    Analytes that show significant changes between healthy and diseased animals in saliva but not in serum. That was the case with biomarkers such as AA, ADA, MPO, AST, LDH, and Ca. The increase in AA could be due to the fact that it is more sensitive to detect the adrenergic stimulation of the salivary gland, whereas, in the case of serum amylase is influenced by pancreas12. ADA has showed previously a better performance in saliva than serum to detect disease conditions13 as well as MPO14. Further studies should be performed to clarify the divergences between saliva and serum of the other analytes found in our study (AST, LDH and Ca) in diseased pigs.

  2. b.

    Analytes that show significant changes between healthy and diseased animals in both saliva and serum. This dynamic was seen in IgG, CRP, Hp, TP, Ferritin, AOPP, FRA, P and Triglycerides. However, the changes were not similar in all analytes. In the case of IgG, CRP and Hp, they showed increases in both saliva and serum, but this response was higher in saliva (6.8-, 5.9- and 4.4-fold, respectively) than in serum (2.1-, 4.0- and 3.8-fold, respectively), even though AUC values of similar magnitude were observed in both sample types. In the case of TP and Ferritin, significant changes of similar magnitude were found in saliva and serum. In contrast, in the cases of AOPP, FRA and Triglycerides, higher values were observed in saliva whereas lower values were detected in serum of diseased animals compared to healthy ones.

  3. c.

    Analytes that showed changes in serum but not in saliva. This dynamic was only observed in BChE and ALP, which showed lower activity in the serum of diseased animals than in healthy ones, whereas no significant changes were seen in saliva.

  4. d.

    Analytes that showed no changes in either saliva or serum. These were the cases of UA, CK, and gGT. In UA, serum did not provide values over the LLOD of the assay used in this report, in contrast with other species, such as humans, in which the measurement of UA in saliva has been revealed as an indicator of UA serum levels15.

The saliva analytes with higher power to discriminate between healthy and diseased pigs were ADA, IgG, AOPP, LDH, and Urea, with an AUC higher than 0.9. From these analytes, IgG and Urea showed also high AUC in serum. The similar capacity in saliva and serum that showed the cases of IgG and Urea could be explained by their passive diffusion from serum to saliva3,4. Also, both analytes showed the higher correlation between saliva and serum in our report. In the case of ADA, as previously indicated, better performance was seen in saliva than in serum to detect disease conditions5. It is important to point out that ADA had a low correlation between saliva and serum, and other analytes, such as AOPP and LDH, did not correlate. Further studies should be performed to clarify the divergences of AOPP and LDH in saliva and serum in diseased pigs. Overall, these results indicated a higher power of saliva to detect diseases with these three biomarkers.

The potential limitations of our study are that these findings are valid only for the specific disease covered in this report, and future studies should explore the relationships of analytes in serum and saliva in other diseases. Another limitation could be that the farm was not specific pathogen free and was positive for porcine reproductive and respiratory syndrome (PRRS), although the animals were negative for infection at the time of sampling. However, this setting could better reflect the conditions of many farms, as PRRSv is endemically present in almost all countries of the European Union16.

Conclusions

Biomarkers such as BChE, ADA, IgG, CRP, Hp, TP, MPO, Ferritin, AOPP, UA, AST, CK, LDH, ALP and gGT can show different changes in saliva and serum depending on the age of the pigs. Also, in pigs with A. pp infection, AA, ADA, MPO, AST, LDH and Ca showed significant increases only in saliva. Therefore, it is a key point to consider these differences for an accurate interpretation of the results and for an appropriate selection of the biomarkers to be used in each type of fluid. More studies in different diseases in which serum and saliva are compared should be performed, in order to have more knowledge about how biomarkers can change in these two fluids and help in an appropriate selection of them.

Methods

Animals

In this cross-sectional study to evaluate the analytes in saliva and serum, a total of 110 pigs (Sus scrofa domesticus; crossbred of [Landrace x Large White] x Pietrain) from a commercial farm in Spain were included. Pigs were housed in the Jalaebro farm (Zaragoza, Spain) with a density ranging 0.20–0.25 m2 per animal and were equipped with standard feeders and nipple drinkers, with ad libitum access to food and water, in accordance with regulations (Council Directive 2001/88/CE of 23 October 2001 amending Directive 91/630/CEE concerning minimum standards for the protection of pigs). The average temperatures on the farms were between 23.4 and 26.4ºC. The animals were vaccinated against porcine circovirus type 2. The sampled farm was considered as Category II (Positive stable) for PRRSv infection according to the classification proposed by Holtkamp and coworkers17. Additionally, PCR assays were performed in serum of all animals included in the study. For this, serum samples were analyzed by real time PCR for the detection of PRRS viral particles with commercial kit VetMAX™ PRRSV EU & NA 3.0 (Thermo Fischer Life Technology) following the manufacturer’s instructions, (yielding in all cases a negative result for PRRSv infection).

In order to study the effect of age and gender, 3 groups of 30 different healthy animals (15 males and 15 females each) were used. Each group corresponded to three different production stages: T0 (recently weaned pigs with a mean age of 30 days-old); T1 (intermediate nursery phase pigs with a mean age of 55 days-old); and T2 (pigs in the fattening period with a mean age of 115 days-old). To better reflect the farm’s condition, animals from different litters were selected for this study with the help of ear-tagging as a means of identification. None of those animals showed any external clinical sign of illness after physical examination.

Additionally, a fourth group integrated by 20 female pigs of a different barn (mean age of 50 days-old) that had clinical signs consisting of severe dyspnea, coughs, anorexia and weight loss was included to evaluate the analytes in saliva and serum in pigs with disease. After PCR testing (BactoReal® Kit Actinobacillus pleuropneumoniae, ingenetix, Vienna, Austria) in serum from all the pigs from this group, it was confirmed that all they were naturally infected by A. pp. (PCR Cq < 45 according to manufacturer instructions). Sampling was performed always before any treatment was administered. PCR was also performed in lung tissue from four pigs that died in the outbreak due to the disease, giving also a positive result. Serum from pigs of T1, which were used as control animals, were also tested for A. pp. giving negative PCR results.

The animal study protocol was approved by the Institutional Review Board (or Ethics Committee) of the University of Murcia (protocol code A13220196; date of approval 4 March 2021). All methods were performed in accordance with the relevant guidelines and regulations. All experiments involving animals were performed in accordance with the ARRIVE guidelines. The diseased pigs with Actinobacillus pleuropneumoniae were treated by administering marbofloxacin at a dose of 8 mg/Kg of body weight as a single intramuscular injection, accompanied by ketoprofen at a dose of 3 mg/Kg of body weight, also given as a single dose.

Collection of samples

Dr. Antonio González-Bulnes, as responsible of R + D activities at the pig farm Jalaebro, authorizes the use of the samples obtained from the animals for use in the research described in the study, being aware of the procedure for obtaining them. All samples were obtained between 8:00–10:00 am in order to avoid any influence on circadian rhythmicity that can occur if samples are obtained at different hours of the day18. Saliva was collected prior to serum extraction, using Salivette tubes (Sarstedt, Aktiengesellschaft & Co. D-51588 Nümbrecht, Germany) with a polypropylene sponge (Esponja Marina, La Griega E. Koronis, Madrid, Spain) instead the cotton swab. Pigs were allowed to chew the sponge until it was completely moist, and the sponges were put into the tubes. After saliva collection, blood samples were obtained through jugular vein venepuncture using vacuum plain tubes (BD Vacutainer, Franklin Lakes, NJ, USA). All samples were kept refrigerated until their arrival at the laboratory and were centrifuged at 3000 g and 4ºC for 10 min to obtain both serum and saliva supernatant, which were stored at − 80ºC until the analysis was performed. After the centrifugation of the saliva samples, those possible contaminants such as cells or food debris were removed, as previously recommended3.

Analytical measurements

Table 8 summarizes all the analytes measured in serum and saliva according to their usefulness as biomarkers and method of measurement. All measurements have been previously validated according to approved protocols19 and were performed in an Olympus AU600 autoanalyzer (Olympus AU600, Olympus Diagnostica GmbH, Ennis, Ireland), with the exception of CRP in saliva and Hp in serum and saliva that were measured by alphalisa technology, a bed-based immunological technique from PerkinElmer (Shelton, CT, USA) that uses a luminescent oxygen-channeling chemistry20. Appropriate references for each method are indicated in the Table.

Table 8 Analytical methods employed and their lower limits of detection (LLOD).
Full size table

Statistical analysis

Shapiro-Wilk test was performed to assess the normality of the data, revealing non-parametric distributions. Therefore, natural log transformation of data was performed prior to analysis. To compare the changes in the analytes in healthy animals between different production stages, a mixed linear model (MLM) was used in which age, sex and sample type were included as fixed factors, and the individual was considered as a random factor. For MLM analyses, three different covariance structures were tested (first order autoregressive, compound symmetry and unstructured) and that providing the lower Akaike’s criteria was selected. A Bonferroni post hoc correction was applied to control the risk of Type I errors in multiple comparisons. Those fixed factors that were not significant were excluded, and the analyses performed again.

To assess differences due to the A. pp infection, the group of diseased animals was statistically compared with the T1 group since the age between groups was similar. A mixed linear model (MLM) was used for this purpose in which group and sample type were included as fixed factors, and the individual as a random factor, followed by Bonferroni post hoc correction. Receiver operating characteristic (ROC) analyses were performed to assess the usefulness of each analyte to differentiate between groups, calculating the area under the curve (AUC) in each case.

Spearman correlation test was performed to assess the relationship between serum and saliva samples within both diseased and healthy groups. In this case, the following rule was used21: r < 0.1: negligible correlation; 0.10–0.39: weak correlation; 0.40–0.69: moderate correlation; 0.70–0.89: strong correlation; r > 0.90: very strong correlation.

The significance level was set at p < 0.05 in all cases. The analyses were performed using IBM SPSS Statistics, version 28.0.1.1.