We evaluated hemolyzed, bacterially contaminated, and Nobuto filter paper-derived serum, collected from 50 Rocky Mountain elk (Cervus elaphus nelson) in 2017 and 2019, divided into eight treatments to determine antibody retention. Serum was analyzed on Brucella abortus-specific fluorescence polarization assay utilizing plates and tubes. Reference titers and serostatus were compared to serum held at 22 C for 4, 8, 12, and 16 d; frozen clotted blood; blood with 2% and 10% elk rumen content (held for 8 d at 22 C); and serum eluted from Nobuto filter paper. Using Cohen's kappa test of agreement, plate assay serostatus agreement was substantial or outstanding in all treatments. Serostatus agreement was outstanding in all treatments utilizing tubes. The mean change in score (treatment minus reference) showed significant negative bias in serosuspect or seropositive animals in the frozen, 2% rumen, and 10% rumen treatments on the plate assay, and the day 16 and 10% rumen treatments on the tube assay, that could ultimately result in an animal being misclassified into a serosuspect or seronegative category. Serum eluted from Nobuto filter paper produced inconsistent results and is not recommended as an alternative to serum derived from blood. Although the potential for misclassification of animals with low titers exists, analyzing hemolyzed and bacterially contaminated serum from Brucella abortus nonendemic areas can increase sample size and the potential to detect seropositive animals.

Diagnostic samples for wildlife disease are often taken in the field under less than ideal conditions. Blood collection, handling, and transport can affect serum quality and, in turn, analytical results (Tuck et al. 2008). One measure of serum quality is hemolysis (Heireman et al. 2017), which is evident visually (Sawant et al. 2007) and by the presence of free hemoglobin in the serum (Sowemimo-Coker 2002). Hemolysis can occur at many points in the preanalytical process. Temperature at collection and during transport can be a critical factor for blood and serum quality (Sowemimo-Coker 2002). Bacterial contamination (Sowemimo-Coker 2002) and time between sample collection and centrifugation can also increase hemolysis (Giavarina and Lippi 2017).

One option for circumventing rigid blood collection and handling requirements is the use of filter paper to collect blood (Curry et al. 2011). Filter paper simplifies collection, transport, and initial storage of blood samples; it can reduce supply and mailing costs, necessary equipment, and produce a more standard, although suboptimal, sample quality. Filter paper-derived serum has been validated in many assays for wildlife diseases (Curry et al. 2014; Kamps et al. 2015), although its functionality on the fluorescence polarization assay (FPA) has been brought into question (Curry et al. 2011).

The FPA detects the presence of Brucella abortus-specific antibodies by evaluating the intensity of plane-polarized light emitted by a fluorescent-labelled tracer bound to an antigen. Previous studies have examined antibody retention in hemolyzed samples utilizing the enzyme-linked immunosorbent assay (Neumann and Bonistalli 2009; Boadella and Gortázar 2011); however, none has utilized the FPA.

The Wyoming Game and Fish Department's Wildlife Health Laboratory annually conducts B. abortus surveillance utilizing hunter-harvested Rocky Mountain elk (Cervus elaphus nelson). Hunters collect blood from their harvested animal and ship to the laboratory or deposit in field coolers for collection by department personnel. Most samples have some degree of hemolysis and bacterial contamination. The goal of this research was to determine if hemolysis and bacterial contamination interferes with antibody detection in two FPAs and determine if the use of blood filter paper is a useful method of blood collection for these assays.

Blood samples were collected from captive and free-ranging Rocky Mountain elk in Wyoming, US and captive elk in Colorado, US. Captive elk were housed at the Wyoming Game and Fish Department's Thorne Williams Wildlife Research Center, in Sybille Canyon, Wyoming (n=12, seronegative) and the US Department of Agriculture's (USDA) Animal and Plant Health Inspection Services National Wildlife Research Center in Fort Collins, Colorado (n=11, seropositive). Samples were collected from captive elk in the spring and fall of 2017. Free-ranging elk were sampled at the Wyoming Game and Fish Department's Grey's River feedground in Alpine, Wyoming in the winter of 2019 (n=27; eight seropositive, 19 seronegative).

Live elk were sampled using a syringe and 18-ga needle. Blood was collected from euthanized elk by opening a major artery and placing a blood tube in the path of blood flow. Nine, 5-mL tubes (Five-0, MTCBio, Sayreville, New Jersey, USA) were collected from each animal, comprising nine treatments of varying degrees of serum quality (Table 1). One tube was centrifuged (5 C, 25 min at 1,455 × G) on the day of collection (reference); four samples were left at 22 C and centrifuged at days 4, 8, 12, and 16; and one tube was frozen at –20 C for maximum hemolysis. To determine effects of bacterial contamination on antibody survival, two blood tubes, containing 100 µL (2%) and 500 µL (10%) elk rumen content, were filled with blood as previously outlined and centrifuged after 8 d at 22 C. Rumen content was collected from euthanized elk and cultured to confirm bacterial growth before use. Blood taken from the animal immediately after euthanasia or by live sampling was also absorbed onto Nobuto blood filter paper (Advantec Type I Blood Sampling Paper, Toyo Roshi Taishu, Ltd., Tokyo, Japan). In brief, Nobuto strips (four per animal) were immersed in blood, allowing contact of the blood absorption area of the strip (equivalent to 0.04 mL blood). Nobuto strips were then placed into a clean, empty blood tube with ventilation holes or a paper envelope to dry completely. The blood absorption area was then cut into small pieces and eluted with phosphate-buffered saline (0.4 mL per full strip) and left overnight at 4 C to attain a 1:10 serum dilution. All samples were stored at 4 C until analyzed (≤1 wk) and then retained at –20 C.

Table 1

Reference serum and descriptions of eight blood treatments to determine antibody retention on a fluorescence polarization assay utilizing plates and tubes.

Reference serum and descriptions of eight blood treatments to determine antibody retention on a fluorescence polarization assay utilizing plates and tubes.
Reference serum and descriptions of eight blood treatments to determine antibody retention on a fluorescence polarization assay utilizing plates and tubes.

A subset of samples (n=13) was analyzed to determine mean hemoglobin content per treatment (QuantiChrom Hemoglobin Assay Kit, BioAssay Systems, Hayward, California, USA). All samples were analyzed using the FPA utilizing both plates and tubes (Brucella abortus Antibody Test Kit, USDA Product Code 5013.30, Ellie LLC, Germantown, Wisconsin, USA) and following the USDA's Animal and Plant Health Inspection Services National Veterinary Services Laboratory standard operating procedures (Fig. 1). Reagents and serum were warmed to room temperature before use. Microplates used for the FPA plate test were black, 96-well, F-bottom plates (no. 655209, Grenier Bio-One, Monroe, North Carolina, USA). The FPA tube test used borosilicate glass disposable culture tubes (no. 73500-1075, Kimble Chase, Vineland, New Jersey, USA). All serologic tests were performed at the Wildlife Health Laboratory. Samples were evaluated in triplicate.

Figure 1

Testing protocol for fluorescence polarization (FPA) plate and tube assays used to determine Brucella abortus specific antibody retention in reference serum and eight blood treatments.

Figure 1

Testing protocol for fluorescence polarization (FPA) plate and tube assays used to determine Brucella abortus specific antibody retention in reference serum and eight blood treatments.

Close modal

Results for the B. abortus FPA were categorized by delta millipolarization unit (mP) scores, placed into three categories: negative (<10 mP), suspect (≥10 mP to ≤20 mP), and positive (>20 mP). The FPA plate assay is utilized as a screening test and the FPA tube assay as a confirmatory test. For sensitivity, specificity, and mean change in detected titers, suspect animals were categorized with positive animals.

Using a paired t-test, mean hemoglobin content was determined to be significantly different (alpha=0.05) from the reference for all treatments except day 4 (Table 2). Cohen's kappa coefficient (κ; SPSS Statistics for Windows, version 25.0, IBM Corp., Armonk, New York, USA) was used to determine the agreement of serostatus between the treatments and reference. Agreement was considered moderate if κ was 0.41–0.60, substantial if 0.61–0.80, and outstanding if 0.81–1.00 (Landis et al. 1977). Cohen's κ coefficient, sensitivity, and specificity with 95% confidence interval (CI), by treatment, for both the FPA plate and tube assay are shown in Table 3.

Table 2

Mean hemoglobin content (mg/dL) of reference serum (day 0); serum held at 22 C for 4, 8, 12, and 16 d; frozen clotted blood; blood with 2% and 10% elk rumen content (held for 8 d at 22 C); and serum derived from Nobuto filter paper (n=13); P value derived from paired t-test between reference and treatment hemoglobin content values.

Mean hemoglobin content (mg/dL) of reference serum (day 0); serum held at 22 C for 4, 8, 12, and 16 d; frozen clotted blood; blood with 2% and 10% elk rumen content (held for 8 d at 22 C); and serum derived from Nobuto filter paper (n=13); P value derived from paired t-test between reference and treatment hemoglobin content values.
Mean hemoglobin content (mg/dL) of reference serum (day 0); serum held at 22 C for 4, 8, 12, and 16 d; frozen clotted blood; blood with 2% and 10% elk rumen content (held for 8 d at 22 C); and serum derived from Nobuto filter paper (n=13); P value derived from paired t-test between reference and treatment hemoglobin content values.
Table 3

Cohen's kappa coefficient (κ) to determine agreement of Brucella abortus serostatus between reference serum (day 0) and serum held at 22 C for 4, 8, 12, and 16 d; frozen clotted blood; blood with 2% and 10% elk rumen content (held for 8 d at 22 C); utilizing fluorescence polarization plate and tube assays and showing SE of the mean, sensitivity, specificity, and 95% confidence interval (CI). Agreement was considered moderate if κ was 0.41–0.60, substantial if 0.61–0.80, and outstanding if 0.81–1.00.

Cohen's kappa coefficient (κ) to determine agreement of Brucella abortus serostatus between reference serum (day 0) and serum held at 22 C for 4, 8, 12, and 16 d; frozen clotted blood; blood with 2% and 10% elk rumen content (held for 8 d at 22 C); utilizing fluorescence polarization plate and tube assays and showing SE of the mean, sensitivity, specificity, and 95% confidence interval (CI). Agreement was considered moderate if κ was 0.41–0.60, substantial if 0.61–0.80, and outstanding if 0.81–1.00.
Cohen's kappa coefficient (κ) to determine agreement of Brucella abortus serostatus between reference serum (day 0) and serum held at 22 C for 4, 8, 12, and 16 d; frozen clotted blood; blood with 2% and 10% elk rumen content (held for 8 d at 22 C); utilizing fluorescence polarization plate and tube assays and showing SE of the mean, sensitivity, specificity, and 95% confidence interval (CI). Agreement was considered moderate if κ was 0.41–0.60, substantial if 0.61–0.80, and outstanding if 0.81–1.00.

Outstanding agreement of serostatus on the FPA plate assay was seen between treatment and reference in the day 4, day 16, and 2% rumen treatments, while the level of agreement seen in the day 8, day 12, frozen and 10% rumen treatments was substantial. Outstanding agreement in serostatus on the FPA tube assay was seen in all treatments. For the FPA plate test, among all treatments sensitivity ranged from 82.61% to 100% and specificity ranged from 80.77% to 100%. Among all treatments measured by the FPA tube test, sensitivity ranged from 91.67% to 100% and specificity was 100%.

The mean change in score (treatment minus reference) was calculated with SD and a 95% CI (Table 4). The distribution of mean change in score per treatment with 95% CI for the FPA plate and tube tests are shown in Figures 2 and 3, respectively. In Figures 2 and 3, when the CI overlaps with zero, there is no statistical difference between the reference and treatment scores (i.e., no bias in the treatment score). From Figure 2, for the seronegative animals there was a slight negative bias in the treatment scores for days 4, 8, 12, and 16 and a significant positive bias in the treatment scores in the frozen and 10% rumen treatments. A negative bias in the seronegative treatments results in scores being more negative and does not lead to misclassification of serostatus. A positive bias in seronegative animals could result in a misclassification to a serosuspect or seropositive category, but those samples would be retested on the confirmatory FPA tube test. Also, from Figure 2, a statistically significant negative bias was observed for the seropositive animals for the frozen, 2% rumen, and 10% rumen treatments, which could result in animals being misclassified into a serosuspect or seronegative category. Additionally, as observed in Table 4, the SD in treatment scores for these groups was larger than the SD in treatment scores for the day 4–16 treatment groups, suggesting potential issues with reproducibility in the frozen, 2%, and 10% rumen treatments. From Figure 3, in seronegative animals, there is a significant negative bias in the day 16 treatment and a slight significant positive bias in the 2% rumen treatment; however, this positive bias could only result in an animal being misclassified into the serosuspect category. Again, in Figure 3, in seropositive animals there is a significant negative bias in the day 16 and 10% rumen treatments. This negative bias could result in an animal being misclassified into a serosuspect or seronegative category. In seropositive animals, an increase in hemolysis also gives more variable results, as shown by the SD.

Table 4

Mean change in score (treatment minus reference), in millipolarization units (mP), with SD and 95% confidence interval (CI), of seven treatments consisting of serum held at 22 C for 4, 8, 12, and 16 d; frozen clotted blood; blood with 2% and 10% elk rumen content (held for 8 d at 22 C) per fluorescence polarization assay (FPA) and serostatus.

Mean change in score (treatment minus reference), in millipolarization units (mP), with SD and 95% confidence interval (CI), of seven treatments consisting of serum held at 22 C for 4, 8, 12, and 16 d; frozen clotted blood; blood with 2% and 10% elk rumen content (held for 8 d at 22 C) per fluorescence polarization assay (FPA) and serostatus.
Mean change in score (treatment minus reference), in millipolarization units (mP), with SD and 95% confidence interval (CI), of seven treatments consisting of serum held at 22 C for 4, 8, 12, and 16 d; frozen clotted blood; blood with 2% and 10% elk rumen content (held for 8 d at 22 C) per fluorescence polarization assay (FPA) and serostatus.
Figure 2

Distribution of fluorescence polarization plate assay treatment score minus reference score, in millipolarization units, of treatments consisting of serum held at 22 C for 4, 8, 12, and 16 d; frozen clotted blood; blood with 2% and 10% elk rumen content held for 8 d at 22 C. CI=confidence interval.

Figure 2

Distribution of fluorescence polarization plate assay treatment score minus reference score, in millipolarization units, of treatments consisting of serum held at 22 C for 4, 8, 12, and 16 d; frozen clotted blood; blood with 2% and 10% elk rumen content held for 8 d at 22 C. CI=confidence interval.

Close modal
Figure 3

Distribution of fluorescence polarization tube assay treatment score minus reference score, in millipolarization units, of treatments consisting of serum held at 22 C for 4, 8, 12, and 16 d; frozen clotted blood; blood with 2% and 10% elk rumen content held for 8 d at 22 C. CI=confidence interval.

Figure 3

Distribution of fluorescence polarization tube assay treatment score minus reference score, in millipolarization units, of treatments consisting of serum held at 22 C for 4, 8, 12, and 16 d; frozen clotted blood; blood with 2% and 10% elk rumen content held for 8 d at 22 C. CI=confidence interval.

Close modal

Nobuto strips analyzed at a 1:50 serum dilution on the FPA tube test resulted in inconsistent and variable results. Inconsistent results have been observed when testing Nobuto-derived serum from caribou and possible reasons have been theorized (Curry et al. 2011). Increasing the sample quantity to reach a 1:50 serum dilution also increases the quantity of substances that may be interfering with the assay, causing inconsistencies.

Although collection and storage guidelines (prelaboratory) are provided to the hunter, other circumstances beyond their control likely play a major role in sample condition (e.g., freezing during transport to the laboratory), resulting in varying degrees of serum quality. Previous investigations of B. abortus serologic tests have not assessed titers in suboptimal samples. Our research demonstrates that hemolyzed serum can be utilized on the B. abortus FPA plate and tube assays; however, an increase in hemolysis results in a decrease in antibody titer, possibly due to increased antibody dilution resulting from the ruptured blood cells. Due to the potential for hemolyzed/bacterially contaminated samples to yield false negative results in animals with low titers, caution must be exercised when utilizing these samples for monitoring within an endemic area, as it could result in a lower calculated prevalence. However, utilizing hemolyzed or bacterially contaminated samples for nonendemic surveillance could increase sample size as well as the potential of detecting seropositive animals.

Funding for this work was provided by the Wyoming Game and Fish Department. We thank the US Department of Agriculture's Animal and Plant Health Inspection Services at the National Wildlife Research Center in Fort Collins, Colorado, the Wyoming Game and Fish Department's Veterinary Services Branch, and Pinedale and Jackson regional personnel for assistance with sample collections.

Boadella
 
M,
Gortázar
 
C.
2011
.
Effect of haemolysis and repeated freeze-thawing cycles on wild boar serum antibody testing by ELISA.
BMC Res Notes
4
:
498
.
Curry
 
PS,
Elkin
 
BT,
Campbell
 
M,
Nielsen
 
K,
Hutchins
 
W,
Ribble
 
C,
Kutz
 
SJ.
2011
.
Filter paper blood samples for ELISA detection of Brucella antibodies in caribou.
J Wildl Dis
47
:
12
20
.
Curry
 
PS,
Ribble
 
C,
Sears
 
WC,
Orsel
 
K,
Hutchins
 
W,
Godson
 
D,
Lindsay
 
R,
Dibernardo
 
A,
Campbell
 
M,
Kutz
 
SJ.
2014
.
Blood collected on filter paper for wildlife serology: Evaluating storage and temperature challenges of field collections.
J Wildl Dis
50
:
308
321
.
Giavarina
 
D,
Lippi
 
G.
2017
.
Blood venous sample collection: Recommendations overview and a checklist to improve quality.
Clin Biochem
50
:
568
573
.
Heiriman
 
L,
Van Geel
 
P,
Musger
 
L,
Heylen
 
E,
Uyttenbroeck
 
W,
Mahieu
 
B.
2017
.
Causes, consequences and management of sample hemolysis in the clinical laboratory.
Clin Biochem
50
:
1317
1322
.
Kamps
 
AJ,
Dubay
 
SA,
Langenberg
 
J,
Maes
 
RK.
2015
.
Evaluation of trapper-collected Nobuto filter-paper blood samples for distemper and parvovirus antibody detection in coyotes (Canis latrans) and raccoons (Procyon lotor).
J Wildl Dis
51
:
724
728
.
Landis
 
JR,
Koch
 
GG.
1977
.
The measurement of observer agreement for categorical data.
Biometrics
33
:
159
174
.
Neumann
 
EJ,
Bonistalli
 
KN.
2009
.
Effect of blood sample handling post-collection on Erysipelothrix rhusiopathiae antibody titres.
Vet J
180
:
325
329
.
Sowemimo-Coker
 
SO.
2002
.
Red blood cell hemolysis during processing.
Transfus Med Rev
16
:
46
60
.
Sawant
 
RB,
Jathar
 
SK,
Rajadhyaksha
 
SB,
Kadam
 
PT.
2007
.
Red cell hemolysis during processing and storage.
Asian J Transfus Sci
1
:
47
51
.
Tuck
 
MK,
Chan
 
DW,
Chia
 
D,
Godwin
 
AK,
Grizzle
 
WE,
Krueger
 
KE,
Rom
 
W,
Sanda
 
M,
Sorbara
 
L,
Stass
 
S,
et al.
2008
.
Standard operating procedures for serum and plasma collection: Early detection research network consensus statement standard operating procedure integration working group.
J Proteome Res
8
:
113
117
.