Adenosine deaminase activity in pig saliva: analytical validation of two spectrophotometric assays

Abstract

We validated 2 assays for the measurement of adenosine deaminase (ADA) activity in the saliva of pigs: the Giusti–Galanti manual method (ADA-GG) and a commercial automated assay (Diazyme Laboratories; ADA-D). Intra-assay coefficients of variation (CVs) were <7 and 9%, and interassay CVs were <12 and 5%, for ADA-GG and ADA-D, respectively. Accuracy was measured by 2 methods: recovery and linearity-under-dilution. Recovery was 82.4–106.8% for ADA-GG, and 92.8–107.9% for ADA-D. Serial dilutions showed R² > 0.95 and 0.99 for ADA-GG and ADA-D, respectively. Linear regression between the methods gave R² = 0.997 (p < 0.0001), and a Bland–Altman plot showed a proportional bias of 112 IU/L (95% confidence interval of −99 to 322 IU/L) for ADA-D. No significant differences were observed between the results obtained by either method in saliva or serum. ADA activity was much higher in porcine saliva than in serum. Salivary ADA activity was significantly higher in lame pigs compared to healthy animals. However, serum ADA activity was significantly lower in lame pigs.

Serum adenosine deaminase (ADA, Enzyme Commission 3.5.4.4) is an enzyme that catalyzes the irreversible conversion of adenosine to inosine and deoxyadenosine to deoxyinosine. These substrates are very toxic to living cells, and ADA plays an important role in detoxification.¹¹ ADA appears in serum and most tissues, especially lymphoid tissues,¹ given that it is necessary for monocyte-to-macrophage differentiation¹² and for the development of B- and T-lymphocytes.¹⁷

ADA activity in serum can increase as a result of leakage of the enzyme from damaged cells.^4,7 In addition, ADA is higher in diseases in which the number of T-lymphocytes increases,^6,10 and hence it is considered a biomarker of cell-mediated immunity.^3,14 In humans, ADA has been proposed as a biomarker of chronic inflammation.¹³

A proteomic study from a group of randomly selected pigs with different pathologic conditions revealed that ADA was among the proteins that were up-regulated in saliva.⁹ Thus, our aim was to validate and compare 2 spectrophotometric assays for ADA determination in pig saliva. We performed analytical validation of the assays and diagnostic validation using healthy pigs and lame pigs, as an example of an inflammatory disease.

Saliva was collected from crossbred growing pigs (Sus scrofa domesticus, Large White × Large White). During lactation, all dams had been vaccinated against Mycoplasma hyopneumoniae (Stellamune Mycoplasma, Pfizer Animal Health, Madrid, Spain) and porcine circovirus 2 (Porcilis PCV, MSD Animal Health, Boxmeer, The Netherlands). Test animals were 2–3-mo-old males in the last phase of fattening and were housed at the Experimental Farm of the University of Murcia (Murcia, Spain). Pigs were given ad libitum access to a nutritionally balanced diet and water. The animals were housed in pens with a minimum space of 0.65 m² per animal (Council of Europe, European Convention for the Protection of Vertebrate Animals Used for Experimental and Other Scientific Purposes, ETS 123. Available from: https://goo.gl/aca2wX) and an average temperature of 23 ± 2°C. Our research protocols were approved by the Bioethical Commission of Murcia University according to the European Council Directives regarding the protection of animals used for experimental purposes (19894).

Saliva was collected from 54 pigs using Salivette tubes (Sarstedt, Nümbrecht, Germany) containing a sponge instead of a cotton swab because sponges were less absorbent and released more saliva following centrifugation. The sampled pigs were allowed to chew the sponge, which was clipped to a flexible thin metal rod, until thoroughly moist, before placing the sponge into the Salivette tube. In addition, serum was obtained from 22 of the sampled animals. Just after saliva collection, pigs were captured with a nose sling and blood collected via jugular venipuncture using tubes without additive (BD Vacutainer, Franklin Lakes, NJ) and allowed to clot. To avoid sample degradation, Salivette and serum tubes were kept in an isothermal box with cold packs until arrival at the laboratory within 4 h of collection. Tubes were centrifuged at 3,500 × g and 4°C for 10 min to obtain saliva and serum.

ADA was measured by 2 different assays. The first assay was based on the method of Giusti and Galanti,⁸ referred to as ADA-GG. This method measures ADA activity through ammonia formation, which is directly proportional to the extinction of indophenol as a final product. First, 100 µL of reagent 1 (21 mM adenosine in 50 mM phosphate buffer, pH 6.5) and 5 µL of sample (S) were incubated at 37°C for 60 min. Then, 300 µL of reagent 2 (106 mM phenol and 0.17 mM sodium nitroprusside) and reagent 3 (11 mM NaOCl and 125 mM NaOH) were added to the reaction medium and vortexed immediately. Incubations were carried out again for 30 min at 37°C. Successively, reaction media were transferred to a 96-well plate and absorbance was measured at 628 nm in a plate reader (PowerWave XS microplate spectrophotometer, BioTek, Winooski, VT, USA). A sample blank (50 mM phosphate buffer, pH 6.5, instead of reagent 1) was made for each sample. A reagent blank (100 µL of 50 mM phosphate buffer, pH 6.5, and 5 µL of water) and an adenosine blank (water instead of sample) for the whole series were also prepared. Ammonium sulfate (0.075 mM) was used as standard instead of reagent 1.

The second assay was a commercial spectrophotometric automated assay (Adenosine deaminase assay kit, Diazyme Laboratories, Poway, CA). This method (referred to as ADA-D) is based on the enzymatic deamination of adenosine to inosine, which is converted to hypoxanthine by purine nucleoside phosphorylase. Hypoxanthine is then converted to uric acid and hydrogen peroxide by xanthine oxidase. Peroxidase is further reacted with N-ethyl-N-(2-hydroxy-3-sulfopropyl)-3-methylaniline and 4-aminoantipyrine in the presence of peroxidase to generate quinine dye, which is kinetically monitored at a 550-nm wavelength.⁷ The method was adapted to an automated analyzer (Olympus AU400, Olympus Diagnostica, Ennis, Ireland) following the manufacturer’s protocol with some modifications.

Because undiluted saliva samples yielded results out of the dynamic range of the method, a 1:8 dilution (1 volume of saliva sample and 7 volumes of distilled water) was applied in all cases and results were multiplied by 8. Dilution was not necessary for serum samples.

Within-run precision, expressed as the coefficient of variation (CV), was calculated by measuring 2 pools of saliva samples containing different levels of ADA 6 times in a single analytical run. In order to obtain 2 pools with different activity, 1 pool was prepared by mixing the same amount of saliva from 5 pigs that yielded high ADA activity measured by ADA-GG and ADA-D methods; the second pool was prepared in the same way using samples from 5 pigs that provided low activity when analyzed by both methods. The same pools were used to determine the between-run precision by analyzing them on 5 consecutive days; the samples were frozen in aliquots, and vials were only thawed as required for each analytical run in order to prevent variation as a result of repeated freeze-thaw cycles. CVs were calculated with the following formula: (standard deviation [SD] of the replicates/arithmetic mean of the replicates) × 100%.

The limit of detection (LOD) was defined as the lowest ADA activity that could be distinguished from a specimen of zero value. It was calculated on the basis of data from 20 replicate determinations of the zero standard (deionized water) as mean value + 2 SD.

Accuracy was indirectly investigated by 2 methods: linearity-under-dilution and recovery. Linearity-under-dilution was determined by using 2 porcine saliva samples with high ADA activity serially diluted (1:2, 1:4, 1:8, 1:16; where 1:2 dilution [1 in 2 dilution] means 1 volume of sample plus 1 volume of diluent, etc.) with deionized water, and the ADA activity was measured by both methods. Because saliva ADA activity was greater than the dynamic range of the method, a 1:8 dilution was applied, and results were corrected before making calculations. The coefficient of determination (R²) was calculated. Intraday recovery was performed by spiking the pools with low and high ADA activities (both 1:8 diluted) with a commercial ADA standard (141.1 IU/L; Diazyme) at different rates. The percentages of recovery were calculated with the following formula: (observed results – unspiked results)/spike amount × 100%. Linearity and recovery experiments were performed by the same person in duplicate, and the arithmetic means of the replicates were used for calculations.

Results obtained with ADA-GG and ADA-D assays in 54 saliva samples were compared by linear regression and Bland–Altman analysis. In addition, 22 paired saliva–serum samples were analyzed with both analytical methods, and correlation between salivary and serum ADA activity was calculated.

To assess the diagnostic validity of ADA measurements, 22 Large White × Large White lame pigs were sampled. An animal was considered lame when it was not able to set any one foot on the ground or to remain standing after routine clinical examination of the animals at the farm. A second group of 32 age- and sex-matched animals with no evidence of clinical disease after physical examination was also sampled as controls. Sera from 13 of the 22 lame pigs and 9 of the 32 healthy pigs were also analyzed.

Mean, SD, CVs, and regression analyses were performed by routine descriptive statistical analysis. Data obtained from healthy and diseased animals were evaluated for normality of distribution, using Shapiro–Wilk and Kolmogorov–Smirnov tests, and were not normally distributed. Data were natural log transformed to assume normal distribution. Linear regression and Bland–Altman plot (difference vs. average) were calculated for comparison between ADA-GG and ADA-D methods after analysis of the 54 saliva samples. Correlation between saliva and serum ADA activities was studied by Spearman correlation assay in 22 paired saliva–serum samples. Two-way analysis of variance (ANOVA) and Bonferroni post-test was used to compare ADA-GG versus ADA-D values between healthy pigs and lame pigs. A value of p < 0.05 indicated significance in all analyses. Data analyses were performed using Excel 2000 (Microsoft, Redmond, WA) and Prism v.5 for Windows (Graph Pad Software, San Diego, CA).

Within-run CVs were <7% for ADA-GG and 9% for ADA-D in both porcine saliva pools (Table 1). Between-run CVs were <12% for ADA-GG and <5% for ADA-D measurements. Linearity gave values of R² > 0.95 (p < 0.01) for the ADA-GG method; however, values below the LOD were obtained for 1:8 and 1:16 dilutions (Fig. 1). For the ADA-D method, regression analyses provided R² > 0.99 (p < 0.001). Recovery was 82–107% for the ADA-GG method, and 93–108% for the ADA-D method (Table 2). LOD was set at 1.14 IU/L for ADA-GG and 0.07 IU/L for ADA-D.

Table 1.

Mean, standard deviation (SD; IU/L), and coefficients of variation (CVs; %) obtained with 2 pools of porcine saliva with different adenosine deaminase activity levels.

Method/Pooled saliva	Within-run			Between-run
	Mean	SD	CV	Mean	SD	CV
ADA-D
High activity	983.1	61.4	6.2	979.7	38.0	3.9
Low activity	423.2	35.2	8.3	423.7	18.0	4.3
ADA-GG
High activity	925.0	64.7	7.0	928.0	38.9	4.2
Low activity	372.1	19.5	5.2	380.3	44.9	11.8

ADA-D = adenosine deaminase (ADA) activity, Diazyme method; ADA-GG = adenosine deaminase (ADA) activity, Giusti–Galanti method.

Figure 1.

Adenosine deaminase (ADA) activity linearity-under-dilution of 2 porcine saliva samples. A. Giusti–Galanti (ADA-GG). B. Diazyme (ADA-D) method. Lower limits of detection were set at 1.14 IU/L for ADA-GG and 0.07 IU/L for ADA-D. ND = no dilution.

Table 2.

Recovery experiment performed by spiking 2 saliva pools with high and low adenosine deaminase (ADA) activity with a commercial ADA standard (141.1 IU/L; Diazyme). Observed results represent the arithmetic mean from 2 replicates.

Method	Pooled saliva with high activity				Pooled saliva with low activity
	Unspiked sample (IU/L)	Spike amount (IU/L)	Observed results (IU/L)	Recovery (%)	Unspiked sample (IU/L)	Spike amount (IU/L)	Observed results (IU/L)	Recovery (%)
ADA-D	98.2	28.2	125.2	95.6	44.2	28.2	74.7	107.9
	73.7	56.4	131.2	101.8	33.2	56.4	85.6	92.9
	61.4	70.6	132.0	100.1	27.7	70.6	98.7	100.8
	49.1	84.7	135.0	101.4	22.1	84.7	105.1	98.0
	24.6	112.9	136.6	99.2	11.1	112.9	117.1	94.0
ADA-GG	92.5	28.2	121.2	101.7	29.4	28.2	59.6	106.8
	69.4	56.4	124.1	97.0	22.1	56.4	68.6	82.5
	57.8	70.6	129.8	102.0	18.4	70.6	82.6	91.0
	46.3	84.7	133.8	103.4	14.7	84.7	97.9	98.3
	23.1	112.9	139.4	103.0	7.4	112.9	112.0	92.7

ADA-D = adenosine deaminase (ADA) activity, Diazyme method; ADA-GG = adenosine deaminase (ADA) activity, Giusti–Galanti method.

Linear regression between salivary ADA-GG and ADA-D showed R² = 0.997 (p < 0.0001), a slope not significantly different from one (1.08 ± 0.008, p < 0.0001), and a y-intercept of 23.1 ± 12.8. The Bland–Altman plot had a proportional bias of 112 IU/L (95% confidence interval −99 to 322 IU/L) between ADA-D and ADA-GG results in saliva (Fig. 2). No significant correlation was observed between saliva and serum for either ADA-GG (Spearman correlation coefficient −0.344, p = 0.062) or ADA-D (Spearman correlation coefficient −0.120, p = 0.593).

Figure 2.

Bland–Altman plot of adenosine deaminase (ADA) activity obtained with Giusti–Galanti (ADA-GG) and Diazyme (ADA-D) methods for 54 saliva samples. For each sample, the y-axis shows the difference in the results obtained between the 2 methods; the x-axis represents the average of the results obtained with the 2 methods.

Results obtained in healthy pigs and pigs with lameness are shown in Figure 3. In saliva, both ADA-GG and ADA-D values were significantly (p < 0.001) higher in lame pigs (median, 25–75th percentiles: 1.6 IU/mL, 1.1–2.2 IU/mL for ADA-GG; 1.9 IU/mL, 1.2–2.4 IU/mL for ADA-D) compared to clinically normal pigs (median, 25–75th percentiles: 0.6 IU/mL, 0.4–1.1 IU/mL for ADA-GG; 0.7 IU/mL, 0.4–1.3 IU/mL for ADA-D). In contrast, ADA-GG and ADA-D was significantly (p < 0.001) lower in serum in lame pigs (median, 25–75th percentiles: 4.6 IU/L, 3.4–6.6 IU/L for ADA-GG; 6.7 IU/L, 4.8–9.4 IU/L for ADA-D ) compared to healthy pigs (median, 25–75th percentiles: 7.4 IU/L, 6.7–9.4 IU/L for ADA-GG; 9.5 IU/L, 8.8–11.2 IU/L for ADA-D). Interaction showed no statistically significant differences between ADA-GG and ADA-D in the response observed between healthy and lame pigs, either in saliva (p = 0.849) or in serum (p = 0.820).

Figure 3.

Adenosine deaminase (ADA) activity in healthy pigs and lame pigs, with Giusti–Galanti (ADA-GG) and Diazyme (ADA-D) methods, in saliva A and serum B. ****p < 0.0001.

The ADA-GG manual assay and the automated ADA-D assay were both able to measure ADA reproducibly in porcine saliva. All saliva samples needed dilution (1:8 or higher) to obtain values within the dynamic range of the assays. The dilution of the saliva samples could have influenced linearity-under-dilution results given that ADA-GG provided values under the LOD of the assay when a 12.5% dilution was applied to 1:8 pre-diluted samples; this was not observed with the ADA-D method. The recovery was 80–120% for both methods, which is within the range accepted for immunoassays.²

High agreement was observed between the methods after linear regression analysis. However, the Bland–Altman plot indicated that, in general, ADA-D provided higher results than ADA-GG, although this bias increased in a proportional way as ADA activity increased. These results could indicate that the ADA-GG method underestimates ADA activity compared to ADA-D. Another major difference between the methods was the time and effort invested for sample analysis. Several homemade reagents were needed for ADA-GG measurement, and the analysis procedure would require >90 min per sample. In addition, adenosine should be freshly prepared.⁸ In contrast, reagents are readily available for ADA-D analysis, the reagents are stable at 4°C until the manufacturer’s expiry date, and analysis was performed in <9 min.

No significant correlation was detected in ADA activity between saliva and serum. Although these data should be taken with caution given the low number of samples analyzed, it would indicate that the source of salivary ADA could be independent to that of serum. We could not find any previous report in which ADA activity was measured in porcine saliva. However, our finding of ADA activity in healthy pigs was much higher than values reported for human saliva, reported as 5.6 ± 3.4 IU/L¹⁵ to 78.5 ± 73.7 IU/L¹⁶ using the ADA-GG method. Conversely, values in serum were similar between the species; mean serum values of 8.6 ± 2.3 U/L⁵ have been reported for healthy humans with the ADA-GG method. Our aim was to validate the assay, rather than to establish reference intervals of ADA for which a larger population would be necessary.

A significant increase was observed in salivary ADA activity with both assays in the lame pigs compared with the healthy pigs. In contrast, serum ADA activity decreased in lame pigs, although these results should be interpreted with caution given the low number of samples tested. In addition, lameness is a generic sign that could appear as the result of various causes, such as infectious arthritis, wounds, or skin lesions, and no specific diagnosis was pursued in our study. In humans, increases in salivary ADA activity have been associated with oral diseases such as squamous cell carcinoma of the tongue,¹⁵ whereas others reported decreases in salivary ADA in patients with oral or laryngeal cancer.¹⁶ ADA activity in serum has been found to be increased in humans with visceral leishmaniasis¹⁹ and in ulcerative colitis, suggesting a partial role of activated T-cell response in the disease pathophysiology.⁵ In contrast, ADA activity in serum has been reported to be decreased in dogs with leishmaniasis.¹⁸ Further studies will be needed to elucidate the mechanism involved in the changes of ADA in saliva and serum found in order to determine the usefulness of salivary ADA as a biomarker for the evaluation of health status, severity of disease, or for monitoring treatment.