Comparison of steroid and thyroid hormone concentrations in blood serum and plasma of captive tigers

Abstract

Analysis of steroid and thyroid hormones is often performed in blood serum. Occasionally though, plasma samples are submitted in lieu of serum for exotic species such as tigers. However, blood tube anticoagulants may affect hormone values. We compared serum and heparin plasma results for 7 hormones in tigers. Serum and plasma samples were collected from 25 tigers and analyzed for progesterone, 17-hydroxyprogesterone, cortisol, androstenedione, testosterone, estradiol, and thyroxine. Using Lin concordance correlation, serum and heparin plasma measures agreed for all hormones except cortisol. However, Passing–Bablok regression only found agreement between serum and heparin plasma measures for androstenedione, testosterone, and estradiol. Median values between the 2 sample types were significantly (p < 0.05) different for progesterone, 17-hydroxyprogesterone, cortisol, and thyroxine. Our results suggest that, for the aforementioned hormones, serum and heparin plasma values may not always be comparable.

Steroid and thyroid hormone values have been described in the literature based on a limited number of tigers^2,5,8,10,12; evaluation of these hormone concentrations has often been performed on blood serum. Interestingly, many of the commercial hormone assay kits used in veterinary diagnostic laboratories are designed for use with human serum or plasma, and therefore, should be evaluated before use in other species. However, assay validation for the occasional sample submitted by zoo veterinarians for non-domesticated species may not be feasible. In our Diagnostic Endocrinology Service at the College of Veterinary Medicine, University of Tennessee (CVM-UT; Knoxville, TN, USA), plasma samples are submitted occasionally, in lieu of serum, from tigers, as well as other exotic species; however, it is not known if hormone concentrations in tiger serum and plasma are comparable.

Anticoagulant additives in blood collection tubes may interfere with certain analytical methods ¹⁷ ; therefore, plasma may not be an appropriate specimen type for all hormone assays. EDTA and heparin are common additives in blood collection tubes used in veterinary medicine. It has been reported that excess EDTA artificially increases canine cortisol concentrations when measured using a solid-phase chemiluminescent enzyme immunoassay ⁴ ; however, it is not known if heparin has a similar effect on cortisol and other steroids. Therefore, our objective was to compare concentrations of progesterone, 17-hydroxyprogesterone (17-OHP), androstenedione, testosterone, estradiol, cortisol, and thyroxine (T4) in blood serum and heparinized plasma of tigers.

We used either archived frozen samples collected in 2016–2018, or samples obtained fresh in 2019 during routine clinical examination, but not sampled specifically for our study. Two tigers had a history of gastrointestinal issues, and one had a submandibular carcinoma; other histories were unremarkable. We used 25 male and female tigers (average age 12.6 y; range: 3–19 y) from a large-cat sanctuary in eastern Tennessee. Each of the tigers was sedated based on an estimated body weight, with target dosages of 0.02 mg/kg of medetomidine HCl (Wildlife Pharmaceuticals), and 0.2 mg/kg of midazolam HCl (NovaPlus; Hospira), followed by 3 mg/kg of ketamine (Ketaset; Fort Dodge). Blood (10 mL) was collected from the medial metatarsal vein and placed into plain and heparinized vacutainer tubes for serum and plasma analyses, respectively. Centrifugation and separation of the serum and heparinized plasma from blood cells were performed and submitted for analysis to the Diagnostic Endocrinology Service at CVM-UT.

Serum and heparin plasma samples were analyzed for progesterone, 17-OHP, androstenedione, testosterone, estradiol, cortisol, and T4. Progesterone, testosterone, cortisol, and T4 concentrations were analyzed using chemiluminescent immunoassays (Immulite 2000; Siemens); 17-OHP, androstenedione, and estradiol concentrations were analyzed using radioimmunoassays (ImmunChem double antibody; MP Biomedicals). To assess the analytical performance of the chemiluminescent immunoassays and radioimmunoassays used in our diagnostic laboratory, within-run precision (or repeatability), within-laboratory precision (or total imprecision), and spike recovery were determined in pooled serum from 4 tigers and pooled heparin plasma from 9 tigers. Tiger samples for pools were selected based on volume available. Five to 10 replicates each day for 3 consecutive days were used to determine within-run and within-laboratory precision. Pooled samples were stored refrigerated between analyses.

A spike-and-recovery test was performed by adding either assay control material (Lyphochek immunoassay plus control; Bio-Rad) for Immulite assays or assay standard material for radioimmunoassays at 1:10 or 1:20 to each hormone pool (Table 1). Final concentrations for each hormone were noted, and percent recovery was calculated by dividing the observed value by the expected value and multiplying by 100%. Ideally, the measured percentage of the added analyte should equal 100 ± 20% of the expected value. Recovery for heparin plasma testosterone and estradiol were outside the ideal measured percentage, suggesting that these assays are not as accurate when measuring tiger heparinized plasma compared to serum.

Table 1.

Within-run precision (or repeatability), within-laboratory precision (or total imprecision) CV and spike recovery for the tested hormones in tiger serum and heparin plasma.

Hormone	Sample type	Pool mean	Within-run (CV%)	Within-laboratory (CV%)	Recovery (%)
Progesterone	S	6.84 nmol/L	11.5	9.1	101
	P	6.23 nmol/L	9.0	2.4	98.0
17-OHP	S	3.70 nmol/L	5.7	7.5	89.2
	P	7.15 nmol/L	8.4	13.1	85.9
Androstenedione	S	2.30 nmol/L	5.9	12.0	99.1
	P	34.4 nmol/L	10.4	11.3	79.4
Testosterone	S	0.98 nmol/L	10.1	11.9	106
	P	1.01 nmol/L	10.1	10.0	69.0
Estradiol	S	189 pmol/L	7.2	15.4	117
	P	327 pmol/L	6.1	9.9	134
Cortisol	S	289 nmol/L	5.2	5.1	105
	P	541 nmol/L	5.5	3.7	102
T4	S	13.5 nmol/L	7.4	7.0	96.8
	P	18.4 nmol/L	9.6	25.3	96.0

17-OHP = 17-hydroxyprogesterone; P = plasma; S = serum; T4 = thyroxine.

Descriptive statistics were performed for each response measure. For hormone values below the lower limit of detection (LLOD), values were substituted using LOD/√2. ¹⁵ The distribution of difference pairs of serum and heparin plasma hormone measures was analyzed for normality using the Shapiro–Wilk test and Q-Q plots. A Shapiro–Wilk test p < 0.05 was considered a violation of normality. Differences between pairs were screened for outliers using box plots. Shapiro–Wilk test results considering the differences between pairs of serum and heparin plasma measures violated the assumption of normality for all measures except cortisol (p = 0.89; Table 2). Additionally, outliers were identified for progesterone (n = 4), 17-OHP (n = 2), androstenedione (n = 1), testosterone (n = 4), estradiol (n = 3), and T4 (n = 4). No outliers were identified for cortisol. Outliers were retained in the dataset and considered in the subsequent analysis.

Table 2.

Hormone concentrations in serum and heparin plasma of tigers.

Hormone*	Serum		Plasma		Serum and plasma pairs
	Mean ± SD	Median (min.–max.)	Mean ± SD	Median (min.–max.)	n (pairs)†	Shapiro–Wilk test
Progesterone	9.23 ± 18.1	2.90‡ (<0.64–88.1)§	8.1 ± 16.2	2.44 (<0.64–79.2)§	24	p < 0.001
17-OHP	3.74 ± 7.72	0.36‡ (<0.30–33.6)§	2.81 ± 6.38	0.21 (<0.30–26.7)§	22	p < 0.001
Androstenedione	12.3 ± 9.8	6.70 (1.01–30.6)	11.5 ± 9.1	7.31 (1.29–31.0)	21	p < 0.001
Testosterone	2.41 ± 4.11	0.63 (<0.12–16.7)§	2.26 ± 3.74	0.70 (<0.12–14.8)§	24	p < 0.001
Estradiol	249 ± 157	205 (87–863)	232 ± 137	200 (87–804)	21	p < 0.001
Cortisol	425 ± 220	408‡ (33–908)	274 ± 174	238 (26–819)	24	p = 0.89
T4	8.84 ± 5.86	8.03‡ (<1.54–24.1)§	7.86 ± 4.92	7.30 (<1.54–19.7)§	24	p = 0.005

17-OHP = 17-hydroxyprogesterone; T4 = thyroxine.

^*Units for all hormone concentrations nmol/L, except estradiol, which is pmol/L.

^†No. of paired observations.

^‡Median serum value significantly higher than heparin plasma value (p ≤ 0.01).

^§Values below assay sensitivity.

Because the differences between serum and heparin plasma for all hormones except cortisol were not normally distributed for any hormone, the paired t-test could not be used; instead, we used a nonparametric approach. A Wilcoxon signed-rank test was performed to evaluate each hormone for differences between serum and heparin plasma samples. Agreement between measurements in both samples was evaluated by performing Lin concordance correlation analyses.^6,16 Concordance correlation <0.9 was considered poor agreement, 0.9–0.95 was considered moderate agreement, >0.95–0.99 was considered substantial, and >0.99 was considered near-perfect agreement. ⁷

Passing–Bablok regression was performed to determine if proportional or systematic bias was present between the dependent measure of heparinized plasma and the independent measure of serum for each hormone. The cusum test for linearity was performed for each analysis. A significant deviation from linearity was determined for 17-OHP (p = 0.03). All other hormones were found to have no significant deviation from linearity (p > 0.05 for each). Bootstrap 95% CIs with 1,000 iterations were computed for slope and intercept parameters. Additionally, a Spearman rank correlation coefficient was computed to assess the relationship between serum and heparin plasma results. Commercial statistical software packages (SPSS v. 28, IBM; MedCalc v.20.015) were used for all analyses; p < 0.05 was considered significant.

Significant differences were found between median serum and heparin plasma values for T4 (Z = −2.56, p = 0.009), cortisol (Z = −4.06, p < 0.001), progesterone (Z = −2.87, p = 0.003), and 17-OHP (Z = −3.3, p < 0.001), with serum measurement higher than the median plasma measurement (Table 2). No differences were observed for androstenedione, testosterone, and estradiol. Significant positive correlation between serum and heparin plasma concentration for each tested hormone was determined by the Spearman correlation coefficient (ρ = 0.87–0.97, p < 0.001, for each); the Lin concordance correlation was moderate-to-substantial (ρ_c = 0.95–0.99) for all hormones except cortisol, which was poor (ρ_c = 0.65; Table 3).

Table 3.

Results of Passing–Bablok regression models to evaluate proportional and systematic bias in serum and heparin plasma samples for each hormone.

Hormone	Intercept	Slope	Spearman ρ	Lin ρc
Progesterone	0.14 (0.04, 0.42)	0.82 (0.77, 0.90)	0.972*	0.985§
17-OHP	0.11 (0.06, 0.21)	0.50 (0.00, 0.58)	0.882*	0.949‡
Androstenedione	0.30 (−0.44, 1.14)	0.98 (0.87, 1.06)	0.944*	0.980§
Testosterone	−0.02 (−0.16, 0.06)	0.94 (0.85, 1.07)	0.941*	0.986§
Estradiol	0.47 (−95.0, 36.3)	0.98 (0.77, 1.44)	0.900*	0.950‡
Cortisol	−25 (−122, 24)	0.67 (0.52, 0.87)	0.872*	0.651†
T4	0.13 (−0.27, 1.07)	0.88 (0.80, 0.97)	0.973*	0.950‡

95% CI in parentheses.

17-OHP = 17-hydroxyprogesterone; T4 = thyroxine.

^*Significant at p < 0.001.

^†Poor agreement.

^‡Moderate agreement.

^§Substantial agreement.

Passing–Bablok regression models were evaluated for each hormone pair to determine if 95% bootstrap CIs contain an intercept value of 0 indicating no systematic bias, and a slope CI including 1, indicating no proportional bias (Table 3; Fig. 1). Intercept and slope CIs for androstenedione, testosterone, and estradiol contained 0 and 1 respectively, thereby indicating no proportional or systematic bias. Thus, there is no evidence to conclude that serum and heparin plasma measurement methods are different for androstenedione, testosterone, and estradiol. Proportional bias was determined for cortisol and T4 because the slope CI did not contain a value of 1. Both proportional and systematic bias were determined for progesterone and 17-OHP.

Figure 1.

Passing–Bablok regression models with 95% shaded CIs.

17-OHP = 17-hydroxyprogesterone; T4 = thyroxine.

When performing Passing–Bablok regression and Lin correlation coefficient analysis, we found that serum and heparin plasma hormone measurements for androstenedione, testosterone, and estradiol agree. Differences between serum and heparin plasma measures arose when comparing hormone median values. Four of the 7 hormones that we analyzed had significantly (p < 0.05) higher median serum values compared to median heparin plasma values (Table 2). Interestingly, cortisol, progesterone, 17-OHP, and T4 were determined to have higher median serum concentration compared to heparin plasma; no significant differences in serum and heparin plasma median values were observed for androstenedione, testosterone, and estradiol. The reason why differences between median serum and heparin plasma values were observed for some hormones but not others is not known.

Mean serum progesterone concentration was significantly higher than mean heparin plasma progesterone concentration in a study of canine blood, ¹⁴ which was consistent with spontaneous activation of complement during coagulation. However, spontaneous activation of complement would not explain why some hormones had no differences between serum and heparin plasma. Fibrin in plasma samples is known to physically interfere with antigen–antibody binding, ¹³ which may be an explanation of differences observed between heparin plasma and serum. Furthermore, it has been reported ¹ that heparin can slow some antibody–antigen reaction rates in second antibody systems, which may explain the differences between serum and heparin plasma measures of 17-OHP.

In many reports on tiger hormone concentrations, analyses have been conducted in either serum^10-12 or fecal samples.^8,9 Serum is the sample type preferred by our diagnostic laboratory because processing fecal samples for hormone analysis can be time consuming. It has been our observation (unpublished) that cortisol, progesterone, and testosterone concentrations are higher in EDTA plasma compared to serum in pigs when using commercial competitive chemiluminescent enzyme immunoassays (Immulite 2000; Siemens). The enzyme employed in the aforementioned immunoassays is alkaline phosphatase, and it has been reported that EDTA inactivates alkaline phosphatase activity by chelating zinc atoms, ³ decreasing the ability of alkaline phosphatase to generate light thereby leading to the erroneous conclusion, in a competitive immunoassay, that the sample has a greater concentration of hormone. ⁴ It is not known what effect heparin may have on this methodology.