1827 Analyst, June 1995, VoE. 120

Determination of the Concentrations of the Steroids Estradiol, Progesterone and Testosterone in Bovine Sera: Comparison of Commercial Dissociation Enhanced Lant hanide Fluorescence lmmunoassay Kits With Conventional Radio and Enzyme lmmunoassays

Christopher T. Elliott, Kathyrn S. Francis, Hugh D. Shortt and William J. McCaughey Drug Residue Laboratory, Veterinary Science Division, Department of Agriculture for Northern Ireland, Stoney Road, Belfast, UK BT4 3SD

The performance of three conventional enzyme and radioimmunoassays routinely used to detect residues of anabolic steroids in cattle sera were compared with dissociation enhanced lanthanide fluorescence immunoassay (DELFIA) kits designed for the hospital market. Slight modifications to the kit reagents were required for the analysis of bovine sera. Owing to the large sample volumes used in conventional assays, detection limits were generally better than those obtained with DELFIA kits, however, assay reproducibility was enhanced using the DELFIA technology. Comparison of sera obtained from cattle implanted with anabolic steroids revealed a good correlation between alternate methods (r2 from 0.91 to 0.97). The DELFIA kits offer a faster method for measuring estradiol, progesterone and testosterone with adequate sensitivity and in a safer environment than that encountered using radioimmunoassays. Keywords: Dissociation enhanced lanthanide fluorescence immunoassay; immunoassay; comparison; residues; steroids

Introduction

The EU directive (649/85/EC) requires EU member states to monitor for the legal and illegal use of a variety of veterinary drugs. Amongst these are the anabolic agents which include the naturally occurring steroids progesterone, estradiol and testosterone and their esters. In 1986 use of these steroids either alone or in combinations as growth-promoting implants was banned in the EU, however, they remain in use in some non-EU countries. Owing to the financial benefits resulting from their use, a minority of EU beef producers have continued to utilize these implants.

In order to monitor cattle populations samples (blood) can be taken from animals on farms or at slaughter. Many laboratories within the EU whose task it is to analyse the samples choose either radioimmunoassay (RIA) or enzyme immunoassay (EIA) as the primary analytical screening technique. This has been due, principally, to the high throughput and sensitivity of these methods coupled with their low cost. The major disadvantages of these conventional tests are the need for organic solvents to extract the steroids from samples prior to analysis for both RIA and EIA and the requirement for radioisotopes and scintillation fluids for RIA. The non-isotopic and highly sensitive DELFIA (Wallac, Turku, Finland) kits for the measurement of progesterone,

estradiol and testosterone have been available for the hospital market for a number of years. Unlike their RIA and EIA counterparts sample extractions are not required.

This paper reports the adaptation of these kits to measure the steroids present in bovine serum and compares the kits with long-established in-house RIA and EIA procedures.

Materials and Methods

Reagents

Steroids and their derivatives were obtained from Steraloids (Pawling, NJ, USA). Tritiated steroids were purchased from Amersham International (Little Chalfont, Buckinghamshire, UK). DELFIA kits were obtained from Wallac (Milton Keynes, UK). Newborn calf serum was supplied by Gibco (Paisley, UK). All other reagents were obtained from Sigma (Poole, Dorset, UK).

Source of Sera

Sera for analysis were obtained from both male and female cattle which were born and raised on Government research establishments. Anabolic growth promoters are not used on these premises.

Positive control sera were obtained from three male and three female calves administered with commercial prepara- tions of progesterone-estradiol (Implixa BM, Hoechst , UK) and testosterone-estradiol (Implixa BF), respectively. Sera was taken from these animals over a two-week period after treatment.

Estradiol-containing sera were also obtained as part of an EU Community Reference Laboratory (CRL) inter-labora- tory comparison (RIVM, Bilthoven, The Netherlands).

Antisera

Polyclonal antibodies to oxime derivatives (Steraloids, NJ, USA) of testosterone, progesterone and estradiol were raised to human serum albumin conjugates prepared by the mixed anhydride technique.1 Rabbits were inoculated as in the schedule described by Elliott et al.2.

In-house Assays

Testosterone by E I A Serum samples (0.5 ml) and testosterone-free control sera spiked with a range of testosterone concentrations were added

Publ

ishe

d on

01

Janu

ary

1995

. Dow

nloa

ded

by M

onas

h U

nive

rsity

on

25/1

0/20

14 0

1:24

:54.

View Article Online / Journal Homepage / Table of Contents for this issue

1828 Analyst, June 1995, Vol. 120

in duplicate to clean glass tubes (120 x 16 mm). A volume (2.5 ml) of diethyl ether was added to each tube, then agitated for 3 rnin on a test-tube shaker (HarvardLTE Multishaker, Old- ham, UK). The tubes were allowed to stand for 1 rnin to allow phase separation. The aqueous layer was then frozen by placing the tubes within an aluminium block pre-cooled in liquid nitrogen. The ether layer was then decanted into clean glass tubes (50 X 10 mm) and evaporated to dryness under a stream of nitrogen gas. Into each tube was added 120 pl of a pre-warmed (37 "C) 2 mmol dm-3 acetate buffer pH 7.0 containing 2 mg ml-l bovine serum albumin and a testos- terone-horseradish peroxidase conjugate at a dilution of 1 + 9999 of a 2 mg ml-l stock solution. The tubes were then vortexed for 30 s.

A 100 pl aliquot of extract was then transferred into a well in a microtitre plate coated with a polyclonal anti-testosterone antibody [prepared by incubating the plate (37 "C) with the antibody (100 pI per well) diluted at 1 + 1999 in a 1 mmol dm-3 acetate buffer pH 7.0 for 2 h].

The microtitre plates were incubated and gently shaken for 2 h at 37 "C or overnight at 4 "C. Following incubation, plates were washed in a 0.9% NaCl, and 0.5% Tween 20 solution; 100 p1 of commercial chromagen (TMB/E, Quadratech, Epsom, Surrey, UK) was added to each well. After incubation for 12 min at 37 "C, 25 pl of 2.5 mol dm-3 sulfuric acid was added to each well to terminate colour production. The absorbances of each well (at 405 nm) were determined on a microtitre plate reader (Bio-Tek EL312e, Vermont, USA) and the concentrations present in the samples were calculated by reference to the calibration graph.

Estradiol by RIA (based on the procedure of Medina)3 Sera samples (500 p1) and estradiol-free sera (obtained from steers) were added in duplicate to clean glass tubes (120 x 16 mm) containing 500 p1 of 50 mmol dm-3 phosphate buffer (pH 7.2). The extraction of these samples was identical to that of testosterone (see above) until dry residues were obtained in glass tubes (50 x 10 mm).

A solution of 50 mmol dm-3 phosphate buffer containing 40000 dpm ml-l of tritiated estradiol and a polyclonal anti- estradiol antibody at a dilution of 1 + 19999 was freshly prepared. A volume (500 p1) of this solution was added to each of the samples and standards and four clean tubes (two to ascertain total counts and two to monitor non-specific binding). All tubes were vortexed for 30 s. The tubes were then incubated at room temperature overnight, after which they were placed in an ice-bath. When the temperature of the assay tubes had cooled to 4 "C (approximately 15 min) 200 yl of a 0.5% m/v activated charcoal-0.05% dextran T70 suspen- sion was added to each of the tubes (except the two total count tubes). All tubes were vortexed for 30 s and incubated in an ice-bath for a further 10 min. The tubes were then centrifuged (Coolspin, MSE, UK) for 10 rnin at 2000g. Following centrifugation the supernatants were decanted into clean glass scintillation vials (50 x 25 mm). A 5 ml volume of scintillation fluid (Optiscint Hisafe, LKB Scintillation Products, UK) was added and after vial capping were incubated at 60 "C for 3 h to obtain equilibrium.

The tritium in each vial was measured by means of scintillation counting (Wallac 1219, Turku, Finland) and the concentrations calculated by reference to the standard curve.

Progesterone by RIA (based on the procedure of Abraham et al.)4 Sample extraction and RIA were performed as described for estradiol with the exception that diethyl ether was substituted by petroleum spirit (bp 40-60 "C).

DELFIA Assays

The kits were used according to the manufacturers instruc- tions. These are as follows.

Testosterone Initially the assays were performed using the manufacturers reagents, however, the human serum supplied was replaced with analyte-free bovine serum.

Commercially obtained newborn calf serum (Gibco, UK) was spiked to give equivalent testosterone standards as were present in the kit reagent (0.3-20 ng ml-l). The anti-rabbit IgG microtitration strips were washed once, and 25 p1 of each standard and serum sample were added directly to duplicate wells followed by 100 p1 of tracer solution, diluted 1 + 49 in assay buffer, and 100 pl of antiserum at the same dilution. The strips were then incubated at room temperature for 2 h on a shaker, then washed six times using an automatic DELFIA washer. Enhancement solution (100 p1) was added to each well, the strips were shaken for a further 5 rnin and read on a time-resolved spectrofluorimeter (1234, Wallac, Turku, Fin- land). The testosterone concentrations present in samples were calculated by reference to the standards included in each assay.

Estradiol The estradiol kit was used as recommended by the manufact- urer. After the standards and samples were applied to the microtitre strips the antiserum was added. The strips were incubated for 30 min at room temperature on the shaker, after which 100 pl per well of tracer was added. The strips were shaken for a further 2 h and the plate developed in the same manner as the testosterone assays.

Progesterone The progesterone kit followed the same protocol as the testosterone kit except that it was unnecessary to wash the microtitration strips before use.

Assay validation The six assays were validated in accordance with the guidelines laid down in EU directive 87/410/EEC. DELFIA standards, prepared in human serum and bovine serum spiked with the same concentrations of analyte were compared. EIA, RIA and DELFIA bovine serum curves were then replicated eight times.

Sera from 20 male cattle known to be free from anabolic treatment were assayed for estradiol and progesterone by RIA and DELFIA; and sera from 20 untreated female cattle were assayed by EIA and DELFIA for the presence of testos- terone. The limits of detection were calculated as being the mean concentration of analyte detected in the samples plus three standard deviations. Within- and between-assay varia- tions were determined by assaying samples spiked with analyte over a range of concentrations.

Results and Discussion The three DELFIA kits were found to be capable oi producing test results within 2-3 h of the sample being submitted for analysis. This compares with 24 h turn-around times with the in-house assays.

Notable differences in the DELFIA standard curve shapes between the human and bovine matices were observed in testosterone and progesterone and to a much lesser extent with estradiol (Figs. 1-3). Standards prepared in negative

Publ

ishe

d on

01

Janu

ary

1995

. Dow

nloa

ded

by M

onas

h U

nive

rsity

on

25/1

0/20

14 0

1:24

:54.

View Article Online

Analyst, June 1995, Vol. 120 1829

h

8 80 Y

v) c.

a 60 0 c .- $ 40 2

0" 20

1 2 3 Log [testosteroneypg per well

Fig. 1 Comparison of DELFIA testosterone standard curves in negative A, human and B, serum.

-

-

-

-

100 I I h 8 80 Y

v) c.

60 0 c .- $ 40

2 0" 20

0 1 I I I I 2 3 4 5 6

Log [estradiol]/fg per well or tube

Fig. 2 Comparison of DELFIA estradiol standard curves in negative A, human and B, bovine serum.

I I I 2 3 4 5

Log [progesteroneypg per well or tube

Fig. 3 Comparison of DELFIA progesterone standard curves in negative A, human and B, bovine serum.

100 [ 1

0 1 I I -4 1 2 3

Log [testosteroneypg per well Fig. 4 Comparison of testosterone standard curves obtained by A, EIA and B, DELFIA procedures (n = 8).

bovine matrix were used for the remainder of the assay validation work.

Standard curve sensitivities were compared at the 25% decrease in signal on each of the mean curves (Figs. 4-6). The sensitivities per well or tube and per ml of sera for each assay are presented in Table 1. The DELFIA testosterone, estradiol

100

h 80 8 v) c 5 60 0

c .- $ 40

E 8 20 n

m

Fig.

I I I I 2 3 4 5 6

Log [estradiolyfg per well or tube

5 Estradiol standard curves obtained by A, RIA and DELFIA procedures (n = 8).

8 100

580 8

Y

v)

c 60

40

.- al

0

0" 20

0 0 1 2 3 4

Log [progesteroneypg per well or tube Fig. 6 Comparison of progesterone standard curves obtained from A, RIA and B, DELFIA procedures (n = 8).

Table 1 Differential sensitivities of the six immunoassay procedures investigated. The sensitivity was determined at the point where a 25% reduction in signal was achieved

Assay type

Testosterone EIA DELFIA

Estradiol RIA DELFIA

Progesterone RIA DELFIA

Sensitivity

pg per well pg ml- 30 60 16 640 31 62 2 80

87 174 17 680

Table 2 Detection limits for each of the six immunoassays investigated

Assay Analyte tY Pe Testosterone EIA

Estradiol RIA

Progesterone RIA

DELFIA

DELFIA

DELFIA

Mead n pgml-l s

20 21 21 20 61 55 20 3 2 20 1 3 20 26 15 20 32 94

Limit of detection/ pgml-l 84

225 10 10 71

314

Publ

ishe

d on

01

Janu

ary

1995

. Dow

nloa

ded

by M

onas

h U

nive

rsity

on

25/1

0/20

14 0

1:24

:54.

View Article Online

1830 Analyst, June 1995, Vol. 120

and progesterone assays were 2,5 and 15 times more sensitive, respectively, than the corresponding in-house method (per well or tube). However, the larger initial volume of sample (500 pl as opposed to 25 pl) added to the in-house tests allowed these assays to give better over-all sensitivities when calcu- lated as pg ml-l of analyte.

The limits of detection (mean of untreated population plus 3s) for each assay are summarized in Table 2. Both testos- terone and progesterone DELFIA methods had higher thresholds than their EINRIA equivalents. The two estradiol tests had identical limits.

The inter- and intra-assay variations for all methods (Tables 3 and 4) show that the DELFIA procedures usually have lower relative standard deviations than the in-house methods, this was reflected by significantly lower variances being found for

Table 3 Intra-assay variations obtained for the six assay procedures investigated over a range of concentrations

Assay Anal yte tY Pe Testosterone EIA

Testosterone DELFIA

Estradiol RIA

Estradiol DELFIA

Progesterone RIA

Progesterone DELFIA

n 10 10 10 10 10 10 10 10 10 10 10 10

Expected Pg ml 200

2000 200

2000 50

200 50

200 500

lo00 500

lo00

Observed Pg ml 195.0

1985.2* 190.4

2178.3 44.0

198.3* 47.6

170.47 476.9* 987.4 760.3

1130.3*

S

2.3 149.9

1.7 92.0 3.2

13.4 3.5 6.8

52.9 118.3 45.6 62.4

* Significantly more accurate than alternate assay ( t > 0.05). t Significantly more precise than alternate assay (f > 0.05).

s r

11.8 7.5 9.1 4.2 7.3 6.8 7.4 4.0

11.1 11.9 6.0 5.5

(Yo 1

Table 4 Inter-assay variations obtained for six assay procedures investigated over a range of concentrations

Assay Expected Observed Sr Analyte type n pgml-1 pgml-* s (Yo) Testosterone EIA 7 200

7 2000 Testosterone DELFIA 7 200

7 2000 Estradiol RIA 7 50

7 200 Estradiol DELFIA 7 50

7 200

7 1000 Progesterone DELFIA 7 500

7 lo00

Progesterone RIA 7 500

197.2* 2087.7

173.6t 1934.6

47.4 225.9 50.3

198.4 467.4

1114.5 485.6

1066.7t

23.2 11.8 161.0 7.7 13.7 7.9

112.2 5.8 6.7 14.1 8.4 3.7 6.8 13.5

11.3 5.7 58.0 12.4

130.8 11.7 46.6 9.6 64.0 6.0

the DELFIAs (f B0.05). Statistical analysis using the students t test showed that the in-house methods were generally more accurate within runs ( t > 0.05), however, few differences were noted in between-run data.

Analysis of samples (n = 18) taken from calves treated with anabolic agents (testosterone-estradiol and progesterone- estradiol) by the in-house and DELFIA methods gave regression coefficients (r2) of 0.93, 0.94 and 0.91 for testos- terone, estradiol and progesterone, respectively. Samples submitted for estradiol assays from the CRL ( n = 4) gave an r2 value of 0.97.

The United Kingdom National Plan requires testing of randomly sampled animals for veterinary drug residues to assess the possibility of drug abuse during production. Assessments are based on a selection of factors including maximum residue limits (MRLs), assay sensitivities and endogenous background levels. For naturally occurring ster- oids the action thresholds have been set at 500 pg ml-1 for progesterone and testosterone and 40 pg ml-l for estradiol. Random samples registering higher than these concentrations are investigated further by on-farm investigations which include collecting additional samples from animals at similar stages of production.

All six of the immunoassays investigated (EIA, RIA and DELFIA) had detection limits below the UK action limits for the natural steroids. In all cases the DELFIA kits had greater sensitivity and better precision profiles than the EINRIA methods. However, the lower sample volume applied to the DELFIA tests rendered their actual detection limits higher with the exception of estradiol. The DELFIA technology, therefore, did not improve the effectiveness of laboratory tests to detect the hormones, however, the efficiencies of test performances were improved.

The major advantages of the kits are that they avoid the need for: solvent extractions, radioactivity, and scintillation cocktails. This leads to faster test completion and a safer working environment. The major disadvantage of the kits was seen as the increased cost of consumable reagents to the laboratory.

References 1 Lewis, L., Elder, P. A., and Barrell, G. K., Steroid Biochem.,

1992,42, 179. 2 Elliott, C. T., McCaughey, W. J., and Shortt, H. D., Food

Additives and Contaminants, 1993, 10, 231. 3 Medina, M. B., J. Agric. Food Chem., 1986,34, 1046. 4 Abraham, G. E., Swerdloff, R., Tulchinsky, P., Odell, W. D.,

J . Clin. Endocrinol, 1971, 32, 619.

Paper 5/00157I Received January 10, 1995

Accepted February 10, 1995 * Significantly more accurate than alternate assay (t > 0.05). t Significantly more precise than alternate assay (f > 0.05).

Publ

ishe

d on

01

Janu

ary

1995

. Dow

nloa

ded

by M

onas

h U

nive

rsity

on

25/1

0/20

14 0

1:24

:54.

View Article Online