In all cases, saliva was obtained previous to blood in order to avoid any possible influence of stress associated with blood collection on the saliva results. Saliva was collected using Salivette tubes (Sarstedt, Aktiengesellschaft & Co. D-51588 Nümbrecht, Germany) containing a sponge (Esponja Marina, La Griega E. Koronis, Madrid, Spain) instead of a cotton swab. The animals were allowed to chew the sponge until thoroughly moist with the help of a flexible thin metal rod. Then, the sponge was placed into the Salivette tube. Venous blood was obtained from venipuncture of the jugular (dogs, horses and pigs) or caudal (cows) veins, using tubes without additive (BD Vacutainer, Franklin Lakes, NJ, USA) and allowed to clot. All samples were kept in ice until arrival at the laboratory for processing (less than 2 hours).
Once at the laboratory, all saliva samples were visually checked and no reddish samples indicating blood contamination were included in the study. The saliva samples were centrifuged (Universal 320R, Hettichzentrifugen, Tuttlingen, Germany) at 3000 x g and 4º C for 10 min. Then, the supernatant was collected in plastic tubes of 1.5mL (Eppendorf), discarding the sediment. Blood tubes were centrifuged in similar conditions than saliva, and serum was collected in Eppendorf. Saliva and serum specimens were stored at -80ºC until analysis.
ADA was analyzed with a commercially available spectrophotometric automated assay (Adenosine Deaminase assay kit, Diazyme Laboratories, Poway, CA, USA). The fundament of this method is as follows: the substrate adenosine is deaminated to inosine by ADA. Inosine is then converted to hypoxanthine by purine nucleoside phosphorylase, which is then converted to uric acid and hydrogen peroxide by xanthine oxidase. The amount of peroxidase produced in the reaction is proportional to the ADA activity in the sample, and it is quantified by reaction with N-ethyl-N-(2-hydroxy-3-sulfopropyl)-3-methylaniline and 4-aminoantipyrine in the presence of peroxidase, leading to a quinine dye kinetically monitored at a 550 nm wavelength . This method was adapted to an automated analyzer (Olympus AU400, Olympus Diagnostica GmbH, Ennis, Ireland) following the manufacturer’s protocol with some modifications for its use in saliva. The lower limit of detection (LLOD) of this method was 0.07 IU/L, following previously published data .
For isoenzyme determinations, the specific ADA1 inhibitor EHNA (Merck KGaA, Darmstadt, Germany) was used. At a proper concentration, EHNA inhibits ADA1 isoenzyme whereas ADA2 remains unaffected . Therefore, tADA and ADA2 isoenzyme can be determined when samples are analyzed in the absence and presence of EHNA, respectively, and the isoenzyme ADA1 is calculated from the difference between both measurements.
Optimization of EHNA concentration for ADA2 measurement in serum and saliva in different species
In order to determine the appropriate concentration of EHNA that should be used in each species for total inhibition of ADA1 isoenzyme, the following samples were used:
Dogs: Samples were obtained from five healthy Beagle dogs (Canis lupus familiaris). All dogs were neutered males, 3.5 ± 0.8 years old and 26.0 ± 7.1 Kg body weight. The animals were located in the Experimental Farm of the University of Murcia (Murcia, Spain).
Horses: Samples were obtained from five healthy horses (Equus caballus), one stallion and four geldings, mean age 10.0 ± 5.1 years old, with body condition score (BCS) 3.4 ± 0.5, including three Spanish horses, one Spanish Arabian and one Warmblood. Horses showed no clinical signs of pain or discomfort after a physical examination.
Pigs: Samples were collected from five apparently healthy growing pigs (Sus scrofa domesticus), Large White x Large White males with 2-3 months-old in the last phase of fattening, housed in the Experimental Farm of the University of Murcia (Murcia, Spain).
Cows: Samples were obtained from five Holstein dairy cows (Bos taurus), lactation 3.5 ± 1.0, mean age 5.3 ± 1.4 years old, days in milk 234.8 ± 9.4, from a commercial dairy herd located in the southeast of Spain. The animals were healthy at physical examination.
Each sample was separated in aliquots. Then, EHNA was added to the saliva and serum samples at increasing concentrations (0.1, 1.0, 4.0 and 8.0mM), whereas an equal volume of diluent was added to one aliquot that was used as control. The proper concentration of EHNA in each species was selected based on its ability to give the same result in ADA value with this concentration than when a higher EHNA concentration was used since this would indicate that at this concentration there is a total inhibition of ADA1.
Development and validation of an automated assay for ADA2 isoenzyme measurement
An automated assay for the measurement of the ADA2 isoenzyme was developed in which EHNA was added to the reagent 1 at an adequate concentration for each species based on the results of the tests described in the previous point. In each species a similar volume of samples obtained from 10 different animals (five with low and five with high ADA2 activity) were mixed in order to prepare two pools of serum and two pools of saliva with different ADA2 activity. Inter-assay imprecision and linearity under dilution were evaluated in serum and saliva samples from the different species by calculating the intra-assay coefficient of variation (CV) and linear regression coefficient, following previously published protocols [22, 28].
In addition, as a part of the validation, the results obtained with the automated procedure were compared with those obtained after the manual addition of the inhibitor. For this approach, the pig was selected as a model because of its high activity in saliva samples. For this purpose, serum and saliva samples with low (N = 15) and high (N = 18) ADA2 activity were obtained. ADA2 isoenzyme was analyzed manually by adding EHNA to the samples and automatically by adding EHNA to the reagent 1, in such concentrations that in both procedures the same final concentration of EHNA in the reaction mixture was achieved.
For this purpose, tADA and its isoenzymes were measured by using the fully automated method in the following samples:
Dog: Samples from 20 dogs were included and divided into two groups. The healthy dog group were integrated by samples of 10 client-owned dogs belonging to the staff of the Animal Medicine and Surgery Department of the University of Murcia. They were 3.9 ± 1.5 years old, with BCS 4.0 ± 1.0, and included three Retrievers, three mixed breed dogs and one of the following breeds: Beagle, French bulldog, Scotish terrier and Brie shepherd. All were neutered males apparently healthy after physical and haematological examinations. The other 10 samples were from client-owned dogs arriving at the Veterinary Teaching Hospital of the University of Murcia, naturally infected with Leishmania infantum with clinical signs. The group with leishmaniosis included three mixed breed dogs and one of the following breeds: Retriever, French bulldog, Collie, Beagle, Irish setter, German shepherd and Rottweiler. There were five males and five females, with 3.0 ± 1.0 years-old, and BCS 2.7 ± 0.5. The clinical signs described in the 10 dogs with leishmaniosis included lymphadenopathy and anaemia (1/10), skin lesions and uveitis (2/10), weight loss and hypoalbuminaemia (3/10), and hyperglobulinaemia (6/10). The diagnoses were based on positive polymerase chain reaction (PCR) and serology results. The concentrations of the acute phase protein ferritin in serum, a biomarker of systemic inflammation in this disease, was analyzed as previously described , ensuring that all healthy individuals had values <190µg/L.
Horse: Samples from 20 horses (10 considered as healthy after physical and blood examinations, and 10 with acute abdominal pain) were included. The healthy animals were male horses admitted for castration or routine health check and included different breeds (seven Spanish horses, one Spanish-Arabian, one Warmblood and one crossbreed), mean age 8.0 ± 4.2 years-old, and BCS 3.5 ± 0.4. They showed no clinical signs of abdominal pain or other diseases during physical examination, as well as haematological or biochemical abnormalities. The group of diseased animals were integrated by horses with acute abdominal disease. This group included animals with different breeds (five Spanish horses, two Warmblood horses, one Lusitanian horse, one Holsteiner and one crossbreed) all males, mean age 11.3 ± 3.3 years-old, and BCS 3.4 ± 0.7. The diagnoses were based on clinical history, physical examination, haematology and plasma biochemistry, transabdominal ultrasonography, rectal examination, nasogastric intubation and on laparotomy findings in surgical cases. The following diagnoses were obtained: three colon impaction with large colon displacement, three stomach impaction, one nephrosplenic entrapment, one impaction of the pelvic flexure, one large colon displacement, and one enteritis. Serum levels of the acute phase protein serum amyloid A (SAA) were measured as a marker of acute systemic inflammation as previously described . All healthy animals showed SAA values <2.3µg/mL.
Pig: Samples from 20 animals (Large White x Large White males with 2-3 months-old in the last phase of fattening) housed in the Experimental Farm of the University of Murcia (Spain) were used. The healthy group was composed of 10 apparently healthy pigs after physical examination at the farm. The diseased group was composed of 10 lame pigs. The presence of lameness was considered based on the observation of the animals according to the scoring system published by Main et al. . An animal was considered lame when it achieved a score ≥1 in the lameness score. Serum C-reactive protein (CRP) concentration was used as a marker of systemic inflammation, and healthy animals gave values <20µg/mL, as previously described .
Cow: Samples from 10 dairy cows (seven Holstein, two Montbellier and one crossbreed), mean age 4.9 ± 1.6 years-old, parity 3.4 ± 1.6 and BCS 3.4 ± 0.7, from a commercial dairy herd located in the southeast of Spain were used. All animals were at the last phase of gestation, apparently healthy and no lameness, mastitis, metritis, ketosis or other health issues were observed. Blood and saliva samples were obtained 13 ± 7 days before calving (Before calving) and at the day of calving (At calving), between January and February of 2019, to avoid any change in the results due to seasonal reasons. The acute phase protein haptoglobin (Hp)  and the total WBC (Advia 120 haematology Analyzer, Siemens Healthcare GmbH, Erlangen, Germany) were used as indicators of inflammation.
Data obtained from ADA measurements were analyzed for normality, giving a non-normal distribution. The changes due to the presence of EHNA at different concentrations were assessed by Friedman’s followed by Dunn’s multiple comparison tests. The concentration that provided statistically significant results with the previous ones but from which there were no more changes was considered as the most appropriate for completely inhibit ADA1 activity. ADA2 results obtained after manual and automated inhibition in the 33 samples (15 with low and 18 with high ADA2 activity) of porcine serum and saliva were compared by linear regression and Bland-Altman plot in which difference between methods were plotted against the average value. Unpaired Mann-Whitney test was used to compare tADA and isoenzymes results between healthy and diseased animals, and Wilcoxon signed rank test was used to compare tADA and isoenzymes results between the two different measurements performed in cows. Spearman correlation coefficients (r) were calculated between ADA results and the biomarkers of inflammation. The correlations were considered according to the r value as very high (≥ 0.90), high (0.70-0.89), moderate (0.50-0.69), low (0.30-0.49), and negligible (< 0.30), following the Rule of Thumb . Data analyses were performed using Excel 2000 (Microsoft Corporation, Redmond, WA, USA) and Graph Pad Software Inc (GraphPad Prism, version 5 for Windows, Graph Pad Software Inc, San Diego, CA, USA). A P value less than 0.05 was considered as significant.