Enzyme-linked immunosorbent assays have been the basis of cytokine and other immunoprotein measurement for several decades. However, these assays can be expensive and laborious and require large sample volumes. The demand for measurement of a constellation of analytes has stimulated development of assays that measure multiple proteins simultaneously, known as multiplex immunoassays.1 Multiplex immunoassays involve traditional immunoassay principles; however, the capture ligands are immobilized in parallel, allowing for identification of multiple proteins concurrently, which traditional ELISAs are unable to provide. Multiplex immunoassays are also more advantageous than traditional ELISAs in that they are less expensive and less labor intensive and require smaller sample volumes.
Use of a multiplex, bead-based cytokine assay system for cats has been reported only once and was limited to evaluation of IFN-γ, IL-10, and IL-12 in tissue culture supernatant.2 A feline-specific, bead-based assay is now commercially available and allows measurement of 19 proteins, including cytokines, chemokines, growth factors, and other immunologically active proteins (CCL2, CCL5, CXCL-1, CXCL-8, CXCL-12, Fas, FLT3LG, GM-CSF, IFN-γ, IL-1β, IL-2, IL-4, IL-6, IL-12p40, IL-13, IL-18, PDGF-BB, SCF, and TNF-α). The purpose of the study reported here was to evaluate this feline-specific multiplex, bead-based assay system for detection of recombinant and native proteins in feline EDTA-treated plasma, heparinized plasma, and serum samples.
Materials and Methods
Cats
Healthy control cats were used for initial assay validation for measurement of feline recombinant and native proteins, to establish expected baseline analyte concentrations, and to assess variability in results over time. For these purposes, healthy cats owned by hospital staff were enrolled with owner consent. Cats were required to be > 1 year of age with no history of illness or medications in the month prior to study inclusion. They were also required to have received a physical examination, CBC, and plasma biochemical analysis, and results of those tests were required to be unremarkable.
To investigate effects of health status on assay results, blood samples from sick cats were obtained. Cats brought to the University of Missouri Veterinary Medical Teaching Hospital and hospitalized in the intensive care unit for at least 24 hours between March and December 2013 were eligible for inclusion in the sick cat group, provided that clinician and owner consent were obtained. Cats receiving chemotherapy or analgesia only were excluded. Data collected on each cat included signalment, diagnosis, whether it survived to hospital discharge, and whether it survived for 28 days after hospitalization (as determined through follow-up phone calls with owners or veterinary reexaminations).
Five purpose-bred specific pathogen–free cats belonging to the University of Missouri were used for measurement of feline native protein concentrations following leukocyte stimulation. The study protocol was approved by the University of Missouri Animal Care and Use Committee.
Blood sample collection
Jugular venous blood samples were collected from all cats by use of 1-inch, 20-gauge needles. Samples were obtained from healthy cats once a day for up to 3 days into plain evacuated tubes,a tubes containing lithium heparin (19 U/mL),b and tubes containing EDTA (4.5 mg/mL).b For the sick cats, 1 jugular venous blood sample was collected into a lithium heparin tube within 24 hours after admission to the intensive care unit. Following blood collection, anticoagulated samples in EDTA tubes were rocked for 1 minute and then cooled. Anticoagulated samples in heparin were immediately placed in ice; blood samples in evacuated tubes were allowed to clot for 30 minutes at room temperature (approx 21°C). All samples were centrifuged at 1,720 × g at 4°C for 10 minutes. Plasma was harvested from EDTA and heparin tubes, and serum was harvested from plain tubes. Harvested portions were stored at −80°C for batch analysis. Whole blood samples collected from purpose-bred cats were prevented from coagulating with sodium heparin and kept at room temperature until processing.
Assay evaluation with feline recombinant proteins
Pooled EDTA-treated and heparinized plasma samples and serum samples from healthy cats were used to evaluate the ability of the feline-specific multiplex, bead-based assayc to detect CCL2, CCL5, CXCL-1, CXCL-8, CXCL-12, Fas, FLT3LG, GM-CSF, IFN-γ, IL-1β, IL-2, IL-4, IL-6, IL-12p40, IL-13, IL-18, PDGF-BB, SCF, and TNF-α. Each of the sample types was spiked with a known amount of each feline recombinant proteind to yield a low, intermediate, and high concentration on the basis of the each assay's dynamic range of detection in accordance with the manufacturer's instructions (Appendix). Spiked pooled plasma and serum samples were added to 6 wells in the same microwell plate to determine the intra-assay CV. The same samples were run on 2 separate plates to determine the interassay CV. Linearity of assay results was assessed by spiking sample solutions with analytes at the high concentration, then comparing results for these neat samples with those of serially diluted samples at a ratio of 1:2, 1:4, and 1:8 via assay diluent.d To assess the effect of heparin on assay performance, blood samples were collected from healthy cats into lithium heparin tubes to yield samples with different heparin concentrations (19, 37, 75, and 150 U/mL). Plasma samples harvested from these different concentrations were spiked with analytes at the high concentration.
The manufacturer provided a serum matrix solutiond that was intended to mitigate the protein matrix effects of high-protein solutions such as serum and plasma. To evaluate the effects of that solution on assay results, assay diluent and pooled serum, heparinized plasma, and EDTA-treated plasma samples were diluted 1:1.5 with serum matrix solution. After dilution, solutions were spiked with recombinant proteins and MFI and percentages of analytes recovered were determined. Extent of analyte recovery from these solutions was determined by comparing the MFI of the spiked test solution with the MFI of 2 control solutions (spiked assay diluent or assay diluent diluted 1:1.5 with serum matrix solution). The following equation was used for this purpose:
Solution protein matrix effects were evaluated by serially diluting pooled plasma or serum samples 1:2, 1:3, 1:4, 1:6, and 1:8, then spiking with a medium concentration of analytes and evaluating recovery. All samples were assayed at least twice. Quality control samples provided by the manufacturer were run on each plate as directed. The LLD was determined by recovery of each analyte from assay diluent (Appendix). Analyte concentrations were measured by use of a multiplexing platform,e and generated data were analyzed with the aid of commercially available analyst software.f Results were interpreted such that a greater MFI equated to a greater analyte concentration.
Assay evaluation with feline native proteins
Individual samples of whole blood obtained from the 5 purpose-bred cats were used to evaluate the ability of the assay to detect native proteins. For this portion of the study, leukocytes in blood samples were stimulated to induce cytokine production by use of a previously described technique for cats.3 Supernatant cytokine concentrations were then measured for in vitro assessment of cytokine production. Briefly, whole blood samples were diluted 1:2 with RPMI complete culture media, transferred to 12-well plates, and stimulated with LPS from Escherichia coli 0127:B8g (100 ng/mL), LTA from Streptococcus faecalisg (1,000 ng/mL), PG from Staphylococcus aureusg (1,000 ng/mL), or control solution (PBSS). Final fluid volume was 2 mL/well. Plates were rocked for 5 minutes and incubated at 37°C in 5% CO2. After 24 hours of incubation, plates were centrifuged (400 × g for 7 minutes) and supernatant was collected. Supernatant samples were stored at −80°C until analysis.
In the second portion of the study in which assay results were obtained for client-owned sick and healthy cats, the assays were performed on heparinized samples obtained on each of the 3 sample-collection days. Variation in analyte concentration among the 3 days was assessed. Analyte concentrations in samples from sick and healthy cats were compared.
Statistical analysis
Statistical analysis was performed by use of statistical software.h Coefficients of variation and percentages of analytes recovered were calculated as described.4,5 For values less than the LLD of each assay, the assay LLD was used for statistical analyses when appropriate. The Shapiro-Wilk test was used to determine whether data were normally distributed. Differences in day-to-day variability in analyte production were assessed by means of 1-way repeated-measures ANOVA. The Mann-Whitney rank sum test was used to compare analyte concentrations between control and sick cats and between sick cats that survived to 28 days after hospitalization and those that did not. Values of P < 0.05 were considered significant.
Results
Cats
One hundred thirty-one cats were admitted to the intensive care unit during the study period and were considered for inclusion in the portion of the study in which concentrations of the 19 immunologically active proteins were measured in blood samples from healthy and sick cats and compared. One hundred one cats were excluded because the clinician or owner did not consent to blood collection. The remaining 30 sick cats were enrolled, consisting of 8 spayed females, 20 castrated males, and 2 sexually intact males. Mean ± SD age was 6.3 ± 4.5 years (range, 1 to 16 years), and mean body weight was 5.3 ± 2.0 kg. Breeds represented were domestic shorthair (n = 24), domestic longhair (3), Maine Coon (2), and Himalayan (1). Cats had been hospitalized because of urinary (n = 8), inflammatory (7), gastrointestinal (5), cardiac (3), metabolic (3), or other (2) diseases or trauma (2). Twenty of the 30 cats remained alive 28 days after hospital discharge.
Nine healthy cats were also enrolled in the study, consisting of 6 castrated males and 3 spayed females. Mean ± SD age was 6.2 ± 4 years (range, 1 to 11 years), and mean body weight was 4.7 ± 0.8 kg. Breeds represented were domestic shorthair (n = 7), Siamese (1), and domestic longhair (1). Blood samples were collected for heparinized plasma harvest from each of these cats, including the sick cats, on 3 consecutive days.
Assay evaluation with feline recombinant proteins
Results for all quality control samples were within the expected ranges as indicated by the manufacturer of the multiplex system for each assay. Intra-assay and interassay CVs for all analytes were consistent with the manufacturer's results (< 10% and 15%, respectively; Table 1). Spike-and-recovery tests performed without serum matrix solution yielded mean percentages of analytes recovered that were less than the target recovery range of 90% to 110% for most analytes. However, mean recovery values were within the target zone for PDGF-BB and TNF-α assays involving heparinized plasma, EDTA-treated plasma, and serum samples when those analytes were measured without addition of serum matrix solution to the samples. Mean recovery values for the remaining 17 analytes were less than those reported by the manufacturer (Table 2).
Mean (range) inter- and intra-assay CVs (%) for a feline-specific multiplex, bead-based assay system used for the measurement of various immunologically active proteins in spiked pooled plasma samples from healthy cats.
| Analyte | Interassay CV | Intra-assay CV |
|---|---|---|
| CCL2 | 8 (1–17) | 5.7 (3–8) |
| CCL5 | 5 (0–11) | 4.0 (3–5) |
| CXCL-1 | 4 (2–9) | 4.4 (3–6) |
| CXCL-8 | 4 (0–12) | 6.6 (2–12) |
| CXCL-12 | 8 (1–15) | 7.2 (6–9) |
| Fas | 2.9 (1–6) | 3.7 (2–5) |
| FLT3LG | 3.9 (0–16) | 4.1 (2–6) |
| GM-CSF | 4 (0–12) | 3.7 (2–5) |
| IFN-γ | 5 (0–12) | 6.2 (3-II) |
| IL-1β | 4 (0–11) | 5.0 (3–7) |
| IL-2 | 14 (2–32) | 6.1 (2–9) |
| IL-4 | 7 (2–21) | 4.2 (3–5) |
| IL-6 | 3 (0–12) | 3.6 (3–4) |
| IL-12p40 | 4 (1–15) | 4.5 (3–7) |
| IL-13 | 9 (1–15) | 4.5 (1–7) |
| IL-18 | 11 (8–16) | 3.4 (2–4) |
| PDGF-BB | 6 (3–10) | 8.3 (4–11) |
| SCF | 5 (2–12) | 4.2 (3–6) |
| TNF-α | 8 (5–11) | 5.0 (5–6) |
Spiked pooled plasma and serum samples were added to the same microwell plate, and that plate was run 6 times to determine the intra-assay CV. The same samples were run on 2 separate plates to determine the interassay CV.
Percentages of analytes recovered from pooled heparinized (19 U/mL) plasma, EDTA-treated (4.5 mg/mL) plasma, and serum samples from healthy cats to which serum matrix solution (SMS) was or was not added, as measured with a feline-specific multiplex, bead-based assay system. (6 samples run in duplicate for each analyte and solution type).
| Calculations based on results for assay diluent | Calculations based on results for assay diluent diluted with SMS | |||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Analyte | Serum | Serum + SMS | Heparinized plasma | Heparinized plasma + SMS | EDTA-treated plasma | EDTA-treated plasma + SMS | Serum | Serum + SMS | Heparinized plasma | Heparanized plasma + SMS | EDTA-treated plasma | EDTA-treated plasma + SMS |
| CCL2 | 48 (38–57) | 50 (40–64) | 65 (37–85) | 15 (12–18) | 35 (0–75) | 23 (16–41) | 50 (41–61) | 52 (44–66) | 306 (93–592) | 74 (42–105) | 280 (38–523) | 95 (66–113) |
| CCL5 | 68 (58–84) | 55 (34–65) | 80 (59–103) | 23 (12–30) | 43 (0–77) | 26 (15–34) | 91 (63–111) | 74 (46–87) | 259 (202–324) | 75 (42–101) | 193 (142–235) | 83 (53–102) |
| CXCL-1 | 13 (3–28) | 14 (4–30) | 5 (1–12) | 5 (2–12) | 4 (0–11) | 6 (3–13) | 252 (125–518) | 23 (7–49) | 18 (5–56) | 24 (8–54) | 22 (12–49) | 27 (11–59) |
| CXCL-8 | 85 (60–99) | 80 (63–93) | 80 (70–83) | 37 (26–58) | 63 (21–80) | 36 (18–57) | 100 (73–121) | 94 (76–113) | 211 (139–255) | 93 (88–101) | 191 (122–239) | 94 (56–108) |
| CXCL-12 | 36 (7–71) | 8 (1–15) | 25 (0–30) | 3 (2–5) | 53 (17–134) | 3 (2–5) | 252 (125–518) | 56 (25–111) | 531 (0–771) | 56 (45–58) | 807 (581–1088) | 54 (29–63) |
| Fas | 52 (31–70) | 45 (39–57) | 43 (26–54) | 19 (13–21) | 35 (11–48) | 20 (15–23) | 79 (47–106) | 68 (45–88) | 152 (105–214) | 68 (54–78) | 143 (88–185) | 74 (62–88) |
| FLT3LG | 59 (30–77) | 49 (8–62) | 54 (28–69) | 15 (0–34) | 43 (21–58) | 27 (4–42) | 78 (40–99) | 64 (11–80) | 203 (96–378) | 46 (0–92) | 148 (103–180) | 86 (13–148) |
| GM-CSF | 22 (0–33) | 27 (9–40) | 24 (11–37) | 12 (4–17) | 20 (3–32) | 13 (6–19) | 34 (12–53) | 42 (16–65) | 90 (50–135) | 48 (25–75) | 95 (64–125) | 52 (33–80) |
| IFN-γ | 56 (27–77) | 16 (8–20) | 57 (36–79) | 9 (5–16) | 43 (17–60) | 10 (6–17) | 249 (139–376) | 72 (40–96) | 430 (357–528) | 70 (51–81) | 375 (196–540) | 80 (66–97) |
| IL-1 | 72 (53–86) | 50 (35–65) | 70 (58–77) | 36 (24–60) | 64 (40–76) | 38 (26–57) | 94 (72–119) | 66 (40–89) | 202 (99–290) | 89 (82–93) | 170 (103–272) | 95 (86–101) |
| IL-2 | 9 (3–15) | 11 (4–17) | 12 (5–27) | 3 (2–6) | 6 (0–17) | 4 (2–6) | 15 (5–26) | 19 (7–30) | 56 (34–95) | 21 (11–34) | 46 (38–60) | 23 (14–36) |
| IL-4 | 69 (42–84) | 46 (33–53) | 61 (44–70) | 19 (15–22) | 51 (17–69) | 21 (16–24) | 111 (65–139) | 73 (52–89) | 239 (194–332) | 75 (65–84) | 230 (173–265) | 83 (71–93) |
| IL-6 | 70 (47–83) | 57 (41–69) | 62 (46–68) | 25 (18–36) | 51 (27–66) | 26 (21–34) | 102 (69–127) | 84 (61–105) | 206 (158–252) | 79 (68–88) | 180 (134–237) | 86 (78–94) |
| IL-12p40 | 25 (0–73) | 19 (0–73) | 23 (0–63) | 18 (1–37) | 10 (0–54) | 16 (0–44) | 27 (0–75) | 20 (0–76) | 77 (0–221) | 37 (3–72) | 27 (0–109) | 35 (0–112) |
| IL-13 | 35 (8–44) | 31 (8–44) | 39 (10–74) | 14 (3–30) | 29 (0–53) | 14 (4–32) | 63 (17–91) | 55 (19–64) | 175 (96–332) | 56 (27–67) | 176 (98–317) | 63 (37–80) |
| IL-18 | 7 (5–13) | 2 (0–4) | 3 (0–8) | 1 (0–3) | 7 (0–21) | 7 (2–26) | 13 (9–25) | 4 (0–8) | 36 (0–143) | 4 (0–11) | 17 (0–30) | 32 (7–129) |
| PDGF-BB | 104 (68–118) | 40 (23–55) | 99 (80–112) | 20 (14–28) | 93 (58–118) | 21 (4–31) | 187 (121–234) | 71 (42–82) | 425 (316–551) | 81 (72–90) | 414 (300–597) | 100 (91–121) |
| SCF | 77 (53–91) | 69 (52–82) | 68 (49–75) | 35 (24–51) | 59 (36–74) | 35 (26–48) | 95 (68–115) | 85 (66–105) | 180 (127–221) | 85 (77–94) | 164 (112–219) | 92 (89–103) |
| TNF-α | 109 (90–120) | 80 (68–95) | 110 (96–124) | 34 (31–37) | 97 (58–114) | 41 (29–60) | 135 (104–167) | 99 (84–121) | 377 (345–503) | 110 (100–122) | 299 (265–331) | 139 (121–238) |
The control solution used for recovery calculations consisted of either spiked assay diluent or assay diluent diluted 1:1.5 with serum matrix solution. Percentage of analyte recovered was calculated by use of the following formula: recovery = (MFI of the test sample – MFI of the background)/(MFI of the control solution – MFI of the background). Notice that although it appears that use of serum matrix solution increased the percentage of analyte recovered, the increase was attributable to a decrease in signal from the assay diluent after addition of serum matrix solution and not to an increase in signal from the test solutions.
In comparisons of recovery results for EDTA-treated plasma, heparinized plasma, and serum samples, the percentage of analyte recovered was greatest for CCL2, CCL5, GM-CSF, IFN-γ, IL-2, IL-13, and TNF-α assays when heparin was used and was greatest for the CXCL-12 assay when EDTA-treated plasma was used. The remaining assays yielded the greatest recovery values when serum samples were used (Table 2). Linearity was difficult to assess because of poor recovery results for many of the assays; the CCL2, CXCL-1, IL-2, and IL-18 assays were not evaluated for linearity for this reason. However, for analytes that yielded sufficient recovery values for assessment, linearity appeared to be reasonable despite the limited detection (Figure 1).
Concentrations (MFIs) of immunologically active proteins in various dilutions of assay diluent (diamonds) and pooled heparinized (19 U/mL) plasma (squares), EDTA-treated (4.5 mg/mL) plasma (triangles), and serum (Xs) samples from healthy cats as measured with a feline-specific multiplex, bead-based assay. Panels represent the following proteins: A, CCL5; B, CXCL-8; C, CXCL-12; D, Fas; E, FLT3LG; F, GM-CSF; G, IFN-γ H, IL-1β; I, IL-4; J, IL-6; K, IL-12p40; L, IL-13; M, PDGF-BB; N, SCF; and O, TNF-α. Sample solutions were spiked with analytes at a high concentration (Table I). Results from this neat sample were then compared with those for samples serially diluted at ratios of 1:2, 1:4, and 1:8. The assay was run to evaluate linearity of results. Tests were run at least twice per sample. A greater MFI indicates that a greater percentage of analyte was recovered. Values for CCL2, CXCL-I, IL-2, and IL-18 assays are not shown because of poor recovery results.
Citation: American Journal of Veterinary Research 77, 5; 10.2460/ajvr.77.5.495
For analyses designed to evaluate effects of heparin concentration on recovery results, data regarding the CCL2, CXCL-1, IL-2, and IL-18 assays were removed because of poor recovery results. A dose-related decrease in the percentage of analyte recovered was identified for all remaining assays as the concentration of heparin increased. When heparin was used at a concentration of 150 U/mL, recovery values decreased from the value achieved with a heparin concentration of 19 U/mL by 18% to 38% for all assays, with the exception of the CXCL-12 assay, the value for which decreased by 83% (Figure 2).
Effects of various concentrations of heparin on results of a feline-specific multiplex, bead-based assay to measure concentrations of immunologically active proteins in pooled plasma samples from healthy cats. Samples were spiked with high concentrations of proteins (Table 1), and tests were run at least twice per sample. See Figure 1 for remainder of key.
Citation: American Journal of Veterinary Research 77, 5; 10.2460/ajvr.77.5.495
Addition of serum matrix solution to the assay diluent, which was intended to improve recovery from solutions containing a high protein concentration, resulted in a decrease in the MFI at several protein concentrations for all assays (Figure 3). When the percentage of analyte recovered was evaluated with assay diluent as the control solution for recovery calculations, values for undiluted (neat) solutions were similar or decreased relative to values for spiked serum, heparinized plasma, and EDTA-treated plasma samples that were not diluted with serum matrix solution. However, when assay diluent diluted with serum matrix solution was used as the control solution for recovery calculations, an increase was identified in recovery values for serum, heparinized plasma, and EDTA-treated plasma samples that were assayed as neat or diluted 1:1.5 with serum matrix solution (Table 2). Thus, diluting serum or plasma samples with serum matrix solution did not result in improvement in signal or recovery results.
Concentrations (MFIs) of immunologically active proteins added at low, medium, and high concentrations to assay diluent (diamonds), assay diluent diluted 1:1.5 with serum matrix (circles), and pooled heparinized (19 U/mL) plasma (squares), EDTA-treated (4.5 mg/mL) plasma (triangles), and serum (Xs) samples from healthy cats as measured with a feline-specific multiplex, bead-based assay. Panels represent the following proteins: A, CCL2; B, CCL5; C, CXCL-1; D, CXCL-8; E, CXCL-12; F, Fas; G, FLT3LG; H, GM-CSF; I, IFN-γ; J, IL-1β;K, IL-2; L, IL-4; M, IL-6; N, IL-12p40; O, IL-13; P, IL-18; Q, PDGF-BB; R, SCF; and S, TNF-α. Concentrations are defined in Table 1 and were based on the dynamic range of detection of the assay in accordance with the assay manufacturer. Tests were run at least twice per sample. A greater MFI indicates that a greater percentage of analyte was recovered. Notice that addition of serum matrix solution to the test medium resulted in a decrease in MFI for all analytes.
Citation: American Journal of Veterinary Research 77, 5; 10.2460/ajvr.77.5.495
For analyses designed to evaluate effects of the sample protein matrix on the percentage of analyte recovered, data regarding the CCL2, CXCL-1, IL-2, and IL-18 assays were removed because of poor recovery results. Heparinized and EDTA-treated plasma samples had a greater MFI prior to dilution than after dilution. The MFI of serum samples was generally lowest prior to dilution. As samples were progressively diluted, the MFI remained relatively fixed for CCL2, CCL5, CXCL-12, FLT3LG, IL-12p40, IL-18, PDGF-BB, and TNF-α assays, indicating a lack of interfering protein matrix effect. The MFI relatively increased with serial dilution of samples in CXCL-1, CXCL-8, Fas, GM-CSF, IFN-γ, IL-1β, IL-4, IL-6, IL-13, and SCF assays, indicating interference by the sample protein matrix (Figure 4).
Concentrations (MFIs) of immunologically active proteins in various dilutions of pooled heparinized plasma (squares), EDTA-treated plasma (triangles), and serum (Xs) samples from 5 healthy cats as measured with a feline-specific multiplex, bead-based assay system. The effect of serum matrix solution on assay results was evaluated by serially diluting pooled plasma or serum samples 1:2, 1:3, 1:4, 1:6, and 1:8 with serum matrix solution supplied by the assay manufacturer, then adding medium concentrations of each analyte (Table 1). See Figure 1 for remainder of key.
Citation: American Journal of Veterinary Research 77, 5; 10.2460/ajvr.77.5.495
Assay evaluation with native proteins
When whole blood samples from 5 purpose-bred cats were stimulated with LTA, LPS, or PG to assess the influence of leukocyte cytokine production on multiplex assay results, no significant difference between samples containing the stimulated leukocytes and samples treated with PBSS as a control solution was identified (data not shown).
When heparinized plasma samples from the 9 healthy cats (1 sample for each of 3 consecutive days) were evaluated by use of the multiplex assay system, no significant differences in day-to-day variability of individual analyte concentrations were identified (Figure 5). Plasma concentrations of the following analytes were lower than the LLD for most samples from healthy cats: CCL2 (15/27), CXCL-1 (26/27), CXCL-8 (17/27), CXCL-12 (22/27), Fas (24/27), GM-CSF (27/27), IL-1β (24/27), IL-2 (27/27), IL-4 (15/27), IL-6 (18/27), IL-13 (27/27), IFN-γ (21/27), PDGF-BB (27/27), and TNF-α (18/27; Figure 6).
Mean concentrations of CCL5 (diamonds), IL-18 (squares), FLT3LG (triangles), IL-12p40 (Xs), and SCF (circles) in heparinized (19 U/mL) plasma samples obtained from 9 healthy cats over a 3-day period as measured with a feline-specific multiplex, bead-based assay system. Values for CCL2, CXCL-1, CXCL-8, CXCL-12, Fas, GM-CSF, IFN-γ, IL-1β, IL-2, IL-4, IL-6, IL-13, pDgF-BB and TNF-α are not shown because concentrations of these analytes in most samples were less than the LLD of the assay. Error bars represent SD.
Citation: American Journal of Veterinary Research 77, 5; 10.2460/ajvr.77.5.495
Concentrations of immunologically active proteins in individual heparinized plasma samples obtained from healthy (n = 9) and sick (30) cats on 3 consecutive days as measured with a feline-specific multiplex, bead-based assay system. Panels represent the following proteins: A, CCL2; B, CCL5; C, CXCL-1; D, CXCL-8; E, FLT3LG; F, IFN-γ; G, IL-1β; H, IL-4; 1, IL-6; J, IL-12p40; K, IL-18; L, SCF; and M, TNF-α. The horizontal line indicates the LLD for each assay. Healthy cats had a significantly (P = 0.028) greater plasma FLT3LG concentration than did sick cats. Values for CXCL-12, Fas, GM-CSF, IL-2, IL-13, and PDGF-BB are not displayed because concentrations were less than the LLD for these analytes for all samples.
Citation: American Journal of Veterinary Research 77, 5; 10.2460/ajvr.77.5.495
For plasma samples from the 30 sick cats, the following analyte concentrations were less than the LLD for > 50% of the cats: CCL2 (20/30), CXCL-1 (25/30), CXCL-8 (17/30), Fas (28/30), GM-CSF (30/30), IFN-γ (26/30), IL-β (25/30), IL-2 (29/30), IL-4 (23/30), IL-6 (21/30), IL-13 (29/30), PDGF-BB (30/30), and TNF-α (22/30). Comparisons of analyte concentrations in heparinized plasma samples between those obtained on day 1 of hospitalization from sick cats (30 samples) and those obtained from healthy cats (9 samples) revealed that healthy cats had a significantly (P = 0.028) greater plasma FLT3LG concentration than did sick cats (Figure 6). No other significant differences were identified between healthy and sick cats or between sick cats that survived to 28 days after hospitalization and those that did not survive (data not shown).
Discussion
The present study was conducted to evaluate a novel feline-specific multiplex, bead-based assay developed for detection of recombinant and native feline cytokines, chemokines, and growth factors in solutions containing low and high protein concentrations. We performed classic spike-and-recovery tests with pooled heparinized plasma, EDTA-treated plasma, and serum samples from healthy cats and feline recombinant proteins, evaluated native proteins by stimulating leukocytes to induce cytokine production, and analyzed heparinized plasma samples from healthy and sick cats. Overall, assays for each of the 19 immunologically active proteins evaluated had good precision collectively but poor recovery results (less than the target zone of 90% to 110%), except for the PDGF-BB and TNF-α assays. Serum samples yielded better recovery results than did EDTA-treated and heparinized plasma samples for most (11/19) of the assays.
Although heparin is an anticoagulant that is known to affect recovery results of immunoassays,6–8 heparinized plasma samples yielded better recovery results than did EDTA-treated plasma in 17 of the 19 assays. However, heparin concentrations in excess of 10 to 30 U/mL affected those results. Addition of serum protein matrix solution to serum or plasma samples did not improve the percentages of analytes recovered. Although poor recovery results limited assessment of assay linearity, most assays yielded a linear change in MFI after dilution that was similar to the linear change identified for analyte-spiked assay diluent. Sample protein matrix negatively impacted recovery results for several assays (CCL2, CCL5, CXCL-12, FLT3LG, IL-12p40, IL-18, PDGF-BB, and TNF-α).
Serum protein matrix solution is provided by the manufacturer of the multiplex assay system and is intended to mitigate the adverse effects of protein matrices on recovery results. Considering results obtained when the control solution was used in recovery calculations, addition of serum protein matrix solution mathematically increased the percentage of analyte recovered. However, it is important to consider the cause of this mathematical increase before use of the serum protein matrix solution becomes standard laboratory practice for assessment of fluids containing high protein concentrations with the evaluated assay. Recovery calculations were based on a comparison of the MFI of the control solution (ie, assay diluent or assay diluent diluted with serum matrix solution) with the MFI of the spiked serum or plasma sample. Presuming that no major background interference existed, an increase in the percentage of analyte recovered could have been a result of either an increase in the MFI of the test sample (ie, the numerator in the calculation) or a decrease in the MFI of the control solution (ie, the denominator). A true increase in recovery values should be mediated through an increase in the MFI of the test sample (ie, improved signal of the test solution, resulting in a larger numerator). When the effect of serum matrix solution on recovery results was evaluated, we noticed that the increase in percentage of analyte recovered was attributable to a decrease in the MFI of the control solution (ie, decrease in the denominator of the equation) and not to an increase in the MFI of the test solution. Therefore, addition of serum matrix solution did not, in reality, lead to an increase in the percentage of analyte recovered or in the signal from the test solution; it simply masked poor recovery results by reducing the MFI of the lower protein concentration in the fluid used for comparison (ie, the denominator of the equation).
To evaluate whether the multiplex assay system could be used to identify native proteins, we induced cytokine production by leukocytes in heparinized whole blood samples from purpose-bred cats by use of a previously established leukocyte cytokine production assay3 and measured concentrations of immunologically active proteins in blood samples from healthy and sick cats. Feline leukocytes produce CXCL-8, IL-1β, and TNF-α, and in other species, CXCL-1, IFN-γ, and IL-6 are produced in response to LPS, LTA, or PG.9,10 For this reason, we expected the assay to identify at least a few of the analytes evaluated. However, there was no significant difference between samples containing the stimulated leukocytes and samples treated with PBSS as a control solution, and most concentrations of most analytes were below the LLD. As expected, most of the healthy cats had circulating analyte concentrations that were lower than the LLD, with no significant day-to-day variability. More than 50% of the samples had a concentration that was lower than the LLD for 14 of the 19 analyte assays for healthy cats and for 15 of those assays for sick cats, limiting our evaluation. Additionally, there were no significant differences in assay results between sick and healthy cats or between sick cats that survived for 28 days after hospital discharge and those that did not survive. The exception was plasma FLT3LG concentration, which was significantly lower in nonsurvivors. Concentrations of circulating analytes change dynamically during the course of disease; therefore, the lack of difference between the sick and healthy cats may simply have been attributable to disease processes, specific activation of immune pathways, or sample collection timing that missed peak plasma concentrations of the analytes. However, given the relatively low recovery results for most of the analyte assays, poor analytical sensitivity could have contributed to the lack of detected analytes in tests of leukocyte cytokine production and in comparisons between sick and healthy cats.
For the multiplex assay system used in the present study, ideal percentages of analytes recovered were 90% to 110%. The overall recovery values for that assay were markedly lower than this over a range of different conditions. Poor recovery results were likely attributable to a matrix effect and poor sensitivity of capture or detection antibodies. Serum and plasma are complex compounds containing proteins, lipids, carbohydrates, electrolytes, and water. Each of these elements has the potential to cause a matrix effect, thereby decreasing the accuracy and sensitivity of ELISAs.11 These effects are more common direct immunoassays, in which the antibodies and enzyme conjugate used are affected by the sample protein matrix.12 Various strategies can be used to minimize matrix effects, including dilution. We observed a progressive increase in recovery as sample protein matrix was diluted for the CXCL-1, CXCL-8, Fas, GM-CSF, IFN-γ, IL-1β, IL-2, IL-4, IL-6, IL-13, and SCF assays, indicating a matrix effect. However, results of CCL2, CCL5, CXCL-12, FLT3LG, IL-12p40, IL-18, PDGF-BB, and TNF-α assays did not appear to be altered by the protein matrix given that serial dilution of the protein-rich fluid did not alter recovery results. Undiluted sample protein matrix yielded greater recovery values when an anticoagulant was present, but after the matrix was initially diluted, recovery values decreased. This phenomenon was not identified for serum samples, suggesting that dilution of the anticoagulant could play a role in the decrease in recovery values that has been reported elsewhere.8,13,14
The manufacturer of the multiplex assay system evaluated in the present study recommends use of EDTA-treated plasma or serum as the test solution; use of heparin as an anticoagulant had not been previously evaluated. Multiple investigators have evaluated the use of several anticoagulants in multiplex assay systems with different results.6–8,15 In the present study, similar or improved recovery results were obtained when heparinized plasma samples were used instead of EDTA-treated plasma or serum samples. However, the effects of heparin on recovery results were variable. Heparin concentrations of 10 to 30 U/mL yielded consistent recovery values, whereas concentrations greater than or less than this concentration range appeared to result in a decrease in the percentage of analyte recovered for most analytes evaluated, highlighting the importance of maintaining heparin concentrations at 10 to 30 U/mL for feline plasma samples.16
Ideally, evaluation of a new assay system should involve comparison to a reference standard. However, almost no feline-specific cytokine assays have been validated for use with solutions containing a high protein concentration. Comparisons were consequently not possible for the study reported here, which was a limitation that should be considered when evaluating the results. In addition, little is known about production of these evaluated analytes in cats; therefore, we were unable to predict which immunologically active proteins we should expect to detect in leukocyte stimulation tests or in samples from sick cats. Additionally, because of the poor assay performance, we did not evaluate the effects of lipemia, icterus, or hemolysis on assay system results.
Overall, poor recovery results and sample protein matrix effects limited the usefulness of the feline-specific multiplex, bead-based assay system for measuring analytes in fluids containing high protein concentrations, such as serum or plasma. Addition of serum matrix solution to test samples did not improve signal capture for the analytes; rather, it reduced the signal in the standard curve used for comparison. Therefore, care should be taken when serum matrix solution is used because it caused an overall decrease in analytic sensitivity of the assays. However, when heparinized plasma was used, CCL2, CCL5, CXCL-8, Ftl-3L, IFN-γ, IL-1β, IL-4, IL-6, PDGF, and SCF assays all yielded recovery values > 50% that were fairly linear, and PDFG-BB and TNF-α assays yielded recovery values in the target range with good linearity. Those particular assays could be useful for detection of high concentrations of analyte in heparinized plasma samples, despite the poor analytic sensitivity in some situations.
Acknowledgments
Some of the MILLIPLEX MAP feline cytokine magnetic bead panel kits and feline recombinant cytokines were provided by EMD Millipore. EMD Millipore did not have a role in the study design; collection, analysis, or interpretation of data; composition or content of the manuscript; or submission of the manuscript for publication.
None of the authors had any personal or financial associations that could have inappropriately influenced or biased the study findings or report contents.
ABBREVIATIONS
| CCL | Chemokine (C-C motif) ligand |
| CXCL | Chemokine (C-X-C motif) ligand |
| CV | Coefficient of variation |
| FLT3LG | FMS-related tyrosine kinase 3 ligand |
| GM-CSF | Granulocyte-macrophage colony stimulating factor |
| IFN | Interferon |
| IL | Interleukin |
| LLD | Lower limit of detection |
| LPS | Lipopolysaccharide |
| LTA | Lipotechoic acid |
| MFI | Mean fluorescence intensity |
| PDGF-BB | Platelet-derived growth factor-BB |
| PG | Peptidoglycan |
| SCF | Stem cell factor |
| TNF | Tumor necrosis factor |
Footnotes
Covidien Ltd, Plymouth, Minn.
Becton, Dickinson and Co, Franklin Lakes, NJ.
MILLIPLEX MAP feline cytokine/chemokine magnetic bead panel, EMD Millipore, Billerica, Mass.
EMD Millipore, Billerica, Mass.
MAGPIX instrument, Luminex Corp, Austin, Tex.
MILLIPLEX Analyst, version 3.5.5, EMD Millipore, Billerica, Mass.
Sigma-Aldrich Corp, St Louis, Mo.
Sigma Plot, version 11.2, Systat Software Inc, Calif.
References
1. Bastarache JA, Koyama T, Wickersham NE, et al. Accuracy and reproducibility of a multiplex immunoassay platform: a validation study. J Immunol Methods 2011; 367: 33–39.
2. Wood BA, O'Halloran KP, Vandewoude S. Development and validation of a multiplex microsphere-based assay for detection of domestic cat (Felis catus) cytokines. Clin Vaccine Immunol 2011; 18: 387–392.
3. Stich AN, DeClue AE. Pathogen associated molecular pattern-induced TNF, IL-1β, IL-6 and CXCL-8 production from feline whole blood culture. Res Vet Sci 2011; 90: 59–63.
4. Chowdhury F, Williams A, Johnson P. Validation and comparison of two multiplex technologies, Luminex and Mesoscale Discovery, for human cytokine profiling. J Immunol Methods 2009; 340: 55–64.
5. Belabani C, Rajasekharan S, Poupon V, et al. A condensed performance-validation strategy for multiplex detection kits used in studies of human clinical samples. J Immunol Methods 2013; 387: 1–10.
6. Biancotto A, Feng X, Langweiler M, et al. Effect of anticoagulants on multiplexed measurement of cytokine/chemokines in healthy subjects. Cytokine 2012; 60: 438–446.
7. Patil R, Shurke S, Paranjape R, et al. Heparin and EDTA anticoagulants differentially affect the plasma and cytokine levels in humans. Scand J Clin Lab Invest 2013; 73: 452–455.
8. Teilmann AC, Kalliokoski O, Jacobsen KR, et al. Impact of heparin and short term anesthesia on the quantification of cytokines in laboratory mouse plasma. Acta Vet Scand 2014; 56: 33–42.
9. Finney SJ, Leaver SK, Evans TW, et al. Differences in lipopolysaccharide- and lipoteichoic acid-induced cytokine/chemokine expression. Intensive Care Med 2012; 38: 324–332.
10. DeClue AE, Nickell J, Chang CH, et al. Upregulation of proinflammatory cytokine production in response to bacterial pathogen-associated molecular patterns in dogs with diabetes mellitus undergoing insulin therapy. J Diabetes Sci Technol 2012; 6: 496–502.
11. Dodig S. Interferences special to immunoassays. Biochem Med 2009; 19: 50–62.
12. Zhang Y, Wang F, Fang L, et al. Rapid determination of ractopamine residues in edible animal products by enzyme-linked immunosorbent assay: development and investigation of matrix effects. J Biomed Biotechnol 2009; 2009: 579175.
13. Hoffmann JN, Hartl WH, Faist E, et al. Tumor necrosis factor measurement and use of different anticoagulants: possible interference in plasma samples and supernatants from endotoxin-stimulated monocytes. Inflamm Res 1997; 46: 342–347.
14. Wong HL, Pfeiffer RM, Fears TR, et al. Reproducibility and correlations of multiplex cytokine levels in asymptomatic persons. Cancer Epidemiol Biomarkers Prev 2008; 17: 3450–3456.
15. Krishnan VV, Ravindran R, Wun T, et al. Multiplexed measurements of immunomodulator levels in peripheral blood of healthy subjects: effects of analytical variables based on anticoagulants, age, and gender. Cytometry B Clin Cytom 2014; 86: 426–435.
16. Adcock DM, Hoefner DM, Kottke-Marchant K, et al. Collection, transport, and processing of blood specimens for testing plasma-based coagulation assays and molecular hemostasis assays; approved guideline—fifth edition. Wayne, Pa: Clinical and Laboratory Standards Institute, 2008;8.
Appendix
Concentrations of immunologically active proteins used in spike and recovery tests and LLDs based on recovery from assay diluent spiked with feline recombinant proteins.
| Concentration (pg/mL) | ||||
|---|---|---|---|---|
| Analyte | Low | Medium | High | LLD (pg/mL) |
| CCL2 | 391 | 3,125 | 25,000 | 406 |
| CCL5 | 8 | 63 | 500 | 1.9 |
| CXCL-1 | 8 | 63 | 500 | 6.4 |
| CXCL-8 | 195 | 1,563 | 1,250 | 9.3 |
| CXCL-12 | 195 | 1,563 | 12,500 | 183 |
| Fas | 39 | 313 | 2,500 | 9.8 |
| FLT3LG | 20 | 156 | 1,250 | 4.7 |
| GM-CSF | 39 | 313 | 2,500 | 36 |
| IFN-γ | 391 | 3,125 | 25,000 | 45.4 |
| IL-1α | 195 | 1,563 | 12,500 | 24.4 |
| IL-2 | 98 | 781 | 6,250 | 37 |
| IL-4 | 39 | 313 | 2,500 | 99 |
| IL-6 | 195 | 1,563 | 1,250 | 47.6 |
| 1L-12p40 | 39 | 313 | 2,500 | 9.6 |
| 1L-13 | 39 | 313 | 2,500 | 10.4 |
| IL-18 | 39 | 313 | 2,500 | 93 |
| PDGF-BB | 781 | 6,250 | 50,000 | 301 |
| SCF | 195 | 1,563 | 12,500 | 46.2 |
| TNF-α | 39 | 313 | 2,500 | 35.9 |
