J. Vet. Med. A 39, 741-746 (1992) 0 1992 Paul Parey Scientific Publishers, Berlin and Hamburg ISSN 0931 - 184X
From the Central Laboratory, Department of Clinical Studies, The Royal Veterinay and Agricultural University, Frederiksberg, Denmark
Adaptation of a Commercially Available Enzyme Linked Immunosorbent Assay (ELISA) for the Determination of Thyroxine in Canine Plasma Samples A. L. JENSEN, R.HQIER and J. S. D. POULSEN Address of authors: Central Laboratory, Department of Clinical Studies, The Royal Veterinary and Agricultural University, Bulowsvej 13, DK-1870 Frederiksberg C, Denmark With 3 tables
(Received for publication june 12, 1992)
Summary The purpose of the present study was to modify a previously evaluated enzyme linked immunosorbent assay (LARSSON and LUMSDEN, 1980) for the determination of the thyroxine (TI) concentration in canine plasma samples. The test kit Enzymun-Test@(BdnngcrMmnhcim) for derermination of T4 concentrations in human plasma was used in connection with the analyzer system Mmnhclm) System ES-33. The assay protocol originally used by LARSSON and Enzymun-Test~(B~hh"ngcr LUMSDEN (1980) was modified by including an additional standard, prolongation of the incubation period for the immunoreaction and the enzymeassay, and reduction of the amount of T4 conjugate. The intra- and interassay coefficient of variations ranged from 3.8 % to 12.5 O/O. The detection limit of the assay was 3.9nmol/l which is well below the T4 concentration observed in plasma from healthy dogs, and the T4 concentrations determined in plasma from healthy dogs before and 3 hours after administration of TRH (38.0 f 7.4 nmol/l and 51.8 k 8.5 nmol/l, respectively) were comparable to those reported in the literature.
Introduction Hypothyroidism is one of the most common canine endocrine diseases. Clinical signs associated with hypothyroidism may include lethargy, bilateral symmetrical alopecia, hyperpigmentation, dry scaly skin, and seborrhea. The classic abnormalities of the haematological and clinical chemical analyses are a normocytic, normochromic, nonand NELSON, 1987). However regenerative anaemia and hypercholesterolaemia (FELDMAN neither the clinical signs nor the classical haematological and clinical chemical abnormalities can be used as the sole criteria when a diagnosis of canine hypothyroidism is to be established. To evaluate the thyroid function in dogs, several methods such as proteinbound iodine determination (BULLOCK,1970) and measurement of endogenous thyroid stimulating hormone (TSH) (LARSSON,1981) have been used. In order to diagnose the presence of canine hypothyroidism the determination of thyroxine (T4) before and after stimulation of the thyroid gland with TSH or thyrotropin releasing hormone (TRH) seems to be the most frequently used method today (FELDMAN and NELSON, 1987; KANEKO, U.S. Copyright Clearance Center Code Statement:
0931 - 184X/92/3910-0741$02.50/0
742
JENSEN,
H~IER and POULSEN
1989). A commonly used technique for the determination of T4 is the radioimmunoassay (RIA) technique, but due to the requirement of relatively expensive instrumentation and most often also of special licensing, this technique may not be an attractive analytical method for some laboratories. Therefore, as an alternative to RIA techniques, an enzyme linked immunosorbent assay (ELISA) may be suggested. Evaluation of an ELISA method for the determination of T4 in canine plasma samples has been reported by LARSSONand LUMSDEN (1980), who concluded that the use of the ELISA method, although being an acceptable alternative, was limited due to unsatisfactory precision and accuracy. The purpose of the present study was therefore to modify the ELISA method and LUMSDEN(1980) in order to optimize the assay, so previously evaluated by LARSSON that the thyroxine concentration in canine plasma samples could be more accurately and precisely measured. Material and Methods Plasma samples The plasma samples used in the present study was isolated from clinically healthy dogs and from hospitalized dogs of different ages and sexes. Blood samples were collected into sodium heparinate vials (VacutainersB, Becton-Dickinson; USA), and plasma fractions were isolated by centrifugation of the blood samples immediately after collection. The plasma samples were stored at -55°C until analysis. Plasma samples which were subjected to repeated analysis were stored at 4°C for maximally one week or at -55°C for maximally three months.
T4 analysis The T4 analysis was performed with the Enzymun-TestBcBoehrInper Mannhem) ELISA kit, which was developed for the determination T4 in human serum and plasma, and the analysis was assisted by the System ES-33. In order to achievea more sensitive assay analyzer system Enzymun-Test~(BoehnngerMannhelm) which fitted better to the T,-concentrations in dogs, the original assay protocol was modified accordingly: The standard series was changed to represent kit standards a, (2 . a + b)/3, b, c, d (0.0, 15.8, 47.4, 96.2, and 167.0nmol/l of T,), and the amount of T4 conjugate was reduced to 5 0 % compared to the original protocol by adding half the amount of stock solution to the supplied buffer. Furthermore, the two incubation steps were prolonged, so that the immunoreaction durated for 60min and the enzymeassay for 60min, in stead of twice times 30min as suggested by the manufacturer. Elsewhere, the original protocol suggested by the manufacturer was followed. The concentrations were recorded in SI-units, i. e. nmol/l of T,. Specz5city The antibodies in the test kit were reported by the manufacturer to crossreact 100 ?Lo with L- and D-T4, 20 % with tetraiodothyroacetic acid, 3.5 Yo and 2.9 Yo with L-T, and D-T,, respectively, and below 0.1 % with 3,5-diiodothyrosine and 3-iodotyrosine. Precision The intraassay variation, expressed as the coefficient of variation (CVYO), was determined from the differences between duplicate determinations at low, medium, and high T4 levels in 22, 24, and 8 canine plasma samples, respectively. The interassay variation was assessed by determining the CV% of 14 to 16 determinations of T4 concentration in the same 2 canine plasma pools in different assays. Accuracy Accuracy was evaluated by investigation of the recovery of known amounts of T4,added as small volumes of the kit’s standards to a canine plasma pool with a low T4-content (13.6 nmol/l). Response to exogeneoxs Tbyrotropin-releasing hormone (TRH) administration Blood samples were collected from 8 clinically healthy male Beagle dogs just before and 3 hours per dog. after intravenous administration of a total dose of 100 pg TRH (Protirelin@(Rmhe))
Adaptation of a Commercially Available Enzyme Linked Immunosorbent Assay
743
The dogs were fasted for 15 hours prior to and during the study. Water was offered ad libitum at all times.
Detection limit The detection limit, i. e. the least detectable concentration (LDC) defined as the smallest single result which, with a stated probability (95 Yo), can be distinguished from the zero standard, was calculated as the concentration corresponding to the absorbancy of the maximum binding minus twice the standard deviation (SD); SD was obtained from an estimate of the pooled variance of 16 replicate absorbance determinations. Statistical methods LDC, intra- and interassay variation (Intra- and interassay CV%) were calculated using routine descriptive statistical procedures (Box et al., 1978). Investigation of recovery of added T4a-as accomplished by regression analysis in linear and in logarithmic transformation. The latter was introduced in order to achieve homogeneity of the variance (WEISBERG, 1985). Lack-of-fit (LOF) tests were performed to investigate whether the data fitted the straight-line regression model appropriately. Briefly, the total error variance was divided into an estimate of the pure error variance according to repeated determinations plus the error variance arosen from lack of fitting to the model. In order to test the model fitting, the two types of error variance was compared by an analysis of variance as described by Box et al. (1978).
Results Precision The intra- and interassay variation are presented in Table 1. The intraassay CV% was found to be 6.3 %, 4.4 %, and 4.4 %, respectively for the difference between 22, 24, and 8 duplicate determinations of plasma samples with a low, medium or high T4 concentration, while the interassay CV% for two canine plasma pools was 3.8 YO and 12.5 YO,respectively.
Accuracy In Table 2 the recovery of T4 added t o a pool of low-T, canine plasma shown as an regression analysis o n the data and as a regression analysis following a logarithmic transformation of the data. It can be seen that the recovery was 100 70 (95 %-confidence limit: 95-102), since the slope of the linear regression was not significantly different from 1 (p = 0.46). The logarithmic regression analysis stated the existence of proportionality between added and recovered T4, since the slope was not different from 1 (p = 0.37). Both regression equations fitted a straight line well following nonsignificant LOF-tests (p = 0.42 for both the linear and the logarithmic model).
Table 1. Intra- and interassay variations of determinations of thyroxine in canine plasma samples
Intraassay'):
Interassayb):
N
Mean f SD nmoVl
CV%
22 24 8 16 7
20.5 f 1.3 40.6 f 1.8 60.1 f 2.6 15.9 f 2.0 96.1 f 3.7
6.3 4.4 4.4 12.5 3.8
Based on duplicate determinations of different plasma samples; b) Based on replicate determination of the same plasma-pool; N Number of plasma samples; Mean Mean of thyroxine concentration; SD Standard deviation; CV% Coefficient of variation.
744
JENSEN,
HOIERand POULSEN
Table2 Linear and logarithmic regression analysis of recovery investigations
Recovery: Linear Logarithmic
model
r
p (slope = 1)
p (LOF)
Y=l3.6+0.986.X Y=-0.183+1.053.X
0.998 0.986
0.46 0.37
0.42 0.42
Linear model: (found) = intercept + slope . (added); Logarithmic model: In(Y) = intercept + slope . (In(X)); r: coefficient of correlation; p (slope): probability for the slope of the model = 1; p (LOF): probability for model fitting (Lack-of-fit test).
Detection limit The least detectable concentration was calculated as 3.9 nmol/l (maximum binding 2.75 AU, standard deviation 0.18 AU), thus depicting LDC below the expected T4 concentrationsinplasmafromhealthydogs of9.0-28.0 nmol/l ( R ~ ~ ~ e t 1978), a l . , 16.7-46.3 nmol/l (FELDMAN and NELSON, 1987), 19.3-39.9nmol/l (KANEKO,1989) and below the median T4 concentrations observed in the clinically healthy dogs of the present study (Table 3).
Response to exogeneous TRH administration Determination of theT4 concentration in plasma samples taken from 8 normal dogs just before and 3 hours following intravenous administration of 100 pgT R H gave the result shown in Table 3. The T4concentration prior to T R H injection was 38.0 k 7.4 nmol/l and it increased to 51.8 k 8.5nmol/l 3 hours after the injection of TRH. The average increment in T4 concentration was 13.8 f 9.2 nmol/l. All dogs exhibited an increment in the TT 4 concentration ranging from 1.05 to 1.96 times the pre-treatment concentrations. Discussion The ELISA method evaluated in the present study showed the following assay characteristics: As displayed in Table 1, the intra- and interassay variation of the ELISA method (4 % - 13 %)were acceptable at normal T4concentrations as well as at low T4concentrations. Furthermore, the modified ELISA method appears be more precise, as judged from the intraand interassay variation, compared to the ELISA method described by LARSSONand LUMSDEN (1980), who reported an intraassay variation of 14.6 % and an interassay variation of 21.3 %. The reason for the increased precision of this modified method could
Table 3. Thyroxine (T4)concentrations before and after exogeneous administration of Thyrotropinreleasing hormone (TRH) in 8 clinically healthy dogs
Mean f SD Range
T, at T=O
T, at T = 3
AT4
Ratio
38.0 f 7.4 (25.3 -47.7)
51.8+ 8.5 (39.7 - 65.9)
13.8 f 9.2 ( 2.6 - 29.2)
1.40 f 0.32 (1.05 - 1.96)
T4atT= 0: ConcentrationofT,(nmoI/l) beforeTRHinjection;T,atT= 3: ConcentrationofT4(nmol/l)3 hours post-injection; AT4: Change in T4 concentration following TRH injection; Ratio: Ratio between T4 concentrations, (T = 3) / (T = 0); Mean: Mean of thyroxine concentration; SD: Standard deviation; Range: Lowest and highest value measured.
Adaptation of a Commercially Available Enzyme Linked Immunosorbent Assay
745
be due to the modifications of the protocol: the reduction of the amount of conjugate makes the dose response curve more sensitive. Besides, the introduction of an additional, low standard (15.8 nmol/l) so that the concentration range (dose range) includes two standards in the range of normally occurring concentrations of T4 in dogs makes the determinations more precise. The prolongation of the incubation times relative to the protocol suggested by the manufacturer and used by LARSSON and LUMSDEN (1980), was introduced in order to ensure equilibrium binding conditions. Still, one should take into consideration that the assay characteristics of the commercial kit might have been changed since the analysis was evaluated by LARSSON and LUMSDEN (1980). The accuracy of the ELISA method was investigated by recovery of T4 added to a pool of canine plasma containing a low concentration of T4 (13.6nmol/l). Recovery of the added T4 was outlined in both a linear regression model, Y = 13.6 + 0.986. X and in a logarithmic regression model, Y = -0.1 83 + 1.053 .X (Table 2). As the slope of the logarithmic regression model was not different from 1, it was shown that proportionality existed between added and recovered T4. Further, from linear regression model, the recovery was shown to be 100 % since the slope of the linear regression line was not different from 1. These findings are in contrast to the reported overestimation of the original ELISA method (LARSSON and LUMSDEN, 1980), thus indicating that the ELISA method used in this study offers a more accurate determination of T4 in canine plasma compared to the original ELISA method. In canine hypothyroidism the' basal plasma concentrations of T4 may be below the T4 concentrations normally observed in plasma from clinically healthy dogs. Thus, it is essential that tke ELISA method is able to determine T4concentrations in plasma samples with T4 concentrations lower than normal. The detection limit for the ELISA method used in this study was calculated as 3.9nmol/l, a T4 concentration well below the reported T4 concentrations in healthy dogs of 9.0-28.0 nmol/l (REAPet al., 1978), 16.7-46.3 nmol/l (FELDMAN and NELSON,1987), 19.3-39.9nmol/l (KANEKO, 1989), and also below the T4 concentrations observed before (25.7-47.7nmol/l) and after (39.7-65.9 nmol/l) TRH administration to the 8 dogs in the present study (Table 3). The T4 concentrations found in 8 normal dogs (Table 3) were 38.0 k 7.4 nmol/l prior to TRH administration. This finding is comparable to the findings of FELDMAN and NELSON (1987) and KANEKO (1989) which reported 16.7-46.3 nmol/l and 19.3-39.9nmoll1, respectively, whereas REAPet al. (1978) found a normal range from 9.0 to 28.0nmol/l. The reported reference intervals are subject to variation, and this fact evidences the necessity for local reference values. In order to diagnose the presence of canine hypothyroidism, determination of T4 before and after stimulation of the thyroid gland with either TSH or TRH is frequently used, the reason for this being that a dog with hypothyroidism will not respond sufficiently to the stimulatiuon (FELDMAN and NELSON, 1987; KANEKO,1989). In Denmark, only the TRH stimulation test is used as TSH is not at present commercially available. However, the protocol for the TRH stimulation test has not been fully established, although the determination of the T4concentration prior to and 4 hours after intravenous administration of 200 pg TRH per dog has been proposed (EVINGER et al., 1985; FELDMAN and NELSON, 1987). Moreover, the value of the TRH stimulation test in separating euthyroid and hypothyroid dogs in clinical situations as well as the effects of non-thyroidal illness and and drug administrations on the response to TRH remains to be studied (FELDMAN NELSON, 1987). However, the results of the TRH stimulation test used in this study (i. e. determination of T4concentration prior to and 3 hours after intravenous administration of 100pg TRH per dog) clearly indicated that the expected increase in T4 concentration following TRH stimulation was recognized with the ELISA method, the ratio between post- and pre-TRH concentrations ranging from 1.05 to 1.96. Further, the T4 concentrations obtained by the TRH stimulation were in agreement with the findings reported by LOTHROPet al. (1984) and EVINGER et al. (1985), who, though using other protocols for the TRH stimulation test, also observed an increase of about 1.5 times the basal T4 concentra-
746
JENSEN, HQIERand POULSEN
tion. Evidently, this indicates that the TRH stimulation test used in this study may be used to evaluate thyroid function. In conclusion, using the ELISA method with the modifications described in this study, the thyroxine (T4)concentration in canine plasma samples can be both precisely and accurately determined compared to the original ELISA method. The detection limit of modified ELISA method is well below the T4 concentration observed in plasma from healthy dogs. Furthermore, the T4concentrations determined in plasma from healthy dogs both before and after performing of TRH stimulation test are in agreement with those reported by others. Acknowledgements The skilful technical assistance of Mrs. V.S. ~ S T V E D Tand Mrs. E.THOMSENis gratefully acknowledged. The financial support of the Danish Agricultural and Veterinary Research Council is also gratefully acknowledged.
References BOX, G.E.P., W.G. HUNTER, and J.S. HUNTER,1978: Statistics for Experimenters. John Wiley & Sons, Inc., New York, USA. BULLOCK, L., 1970: Protein-bound iodine determination as a diagnostic aid for canine hypothyroidism. J. Am. Vet. Med. Assoc. 156, 892-899. J. V., R. W. NELSON,and G. D. BOTTOMS,1985: Thyrotropin-releasing hormone stimulaEVINGER, tion testing in healthy dogs. Am. J. Vet. Res. 46, 1323-1325. FELDMAN, E. C., and R. C. NELSON, 1987: Hypothyroidism. Canine Hypothyroidism. In: Canine and Feline Endocrinology and Reproduction (editor: D. PEDERSEN), W. B. Saunders Company Philadelphia, pp. 55-90. KANEKO, J. J., 1989: Thyroid function. In: Clinical Biochemistry of Domestic Animals, 4th ed. Academic Press Inc. San Diego, pp. 630-649. (editor: J. J. KANEKO), LARSSON, M., and J.H. LUMSDEN, 1980: Evaluation of an enzyme linked immunosorbent assay (ELISA) for the determination of plasma thyroxine in dogs. J. Vet. Med. A 27, 9-15. LARSSON, M., 1981: Evaluation of a human TSH radioimmunoassay as a diagnostic test for canine primary hypothyroidism. Acta Vet. Scand. 22, 589-591. and V. A. FADOK,1984: Canine and feline thyroid function assessment LOTHROP,C. D., P. M. TAMAS, with the thyrotropin-releasing hormone response test. Am. J. Vet. Res. 45, 2310-2313. 1978: Thyroxine and triiodothyronine Levels in Ten Species REAP,M., C. CASS,and D. HIGHTOWER, of Animals. Southwestern Vet. 31, 31-34. S., 1985: Applied Linear Regression. 2nd ed. John Wiley & Sons, Inc., New York, USA. WEISBERG,
From the Central Laboratory, Department of Clinical Studies, The Royal Veterinay and Agricultural University, Frederiksberg, Denmark
Adaptation of a Commercially Available Enzyme Linked Immunosorbent Assay (ELISA) for the Determination of Thyroxine in Canine Plasma Samples A. L. JENSEN, R.HQIER and J. S. D. POULSEN Address of authors: Central Laboratory, Department of Clinical Studies, The Royal Veterinary and Agricultural University, Bulowsvej 13, DK-1870 Frederiksberg C, Denmark With 3 tables
(Received for publication june 12, 1992)
Summary The purpose of the present study was to modify a previously evaluated enzyme linked immunosorbent assay (LARSSON and LUMSDEN, 1980) for the determination of the thyroxine (TI) concentration in canine plasma samples. The test kit Enzymun-Test@(BdnngcrMmnhcim) for derermination of T4 concentrations in human plasma was used in connection with the analyzer system Mmnhclm) System ES-33. The assay protocol originally used by LARSSON and Enzymun-Test~(B~hh"ngcr LUMSDEN (1980) was modified by including an additional standard, prolongation of the incubation period for the immunoreaction and the enzymeassay, and reduction of the amount of T4 conjugate. The intra- and interassay coefficient of variations ranged from 3.8 % to 12.5 O/O. The detection limit of the assay was 3.9nmol/l which is well below the T4 concentration observed in plasma from healthy dogs, and the T4 concentrations determined in plasma from healthy dogs before and 3 hours after administration of TRH (38.0 f 7.4 nmol/l and 51.8 k 8.5 nmol/l, respectively) were comparable to those reported in the literature.
Introduction Hypothyroidism is one of the most common canine endocrine diseases. Clinical signs associated with hypothyroidism may include lethargy, bilateral symmetrical alopecia, hyperpigmentation, dry scaly skin, and seborrhea. The classic abnormalities of the haematological and clinical chemical analyses are a normocytic, normochromic, nonand NELSON, 1987). However regenerative anaemia and hypercholesterolaemia (FELDMAN neither the clinical signs nor the classical haematological and clinical chemical abnormalities can be used as the sole criteria when a diagnosis of canine hypothyroidism is to be established. To evaluate the thyroid function in dogs, several methods such as proteinbound iodine determination (BULLOCK,1970) and measurement of endogenous thyroid stimulating hormone (TSH) (LARSSON,1981) have been used. In order to diagnose the presence of canine hypothyroidism the determination of thyroxine (T4) before and after stimulation of the thyroid gland with TSH or thyrotropin releasing hormone (TRH) seems to be the most frequently used method today (FELDMAN and NELSON, 1987; KANEKO, U.S. Copyright Clearance Center Code Statement:
0931 - 184X/92/3910-0741$02.50/0
742
JENSEN,
H~IER and POULSEN
1989). A commonly used technique for the determination of T4 is the radioimmunoassay (RIA) technique, but due to the requirement of relatively expensive instrumentation and most often also of special licensing, this technique may not be an attractive analytical method for some laboratories. Therefore, as an alternative to RIA techniques, an enzyme linked immunosorbent assay (ELISA) may be suggested. Evaluation of an ELISA method for the determination of T4 in canine plasma samples has been reported by LARSSONand LUMSDEN (1980), who concluded that the use of the ELISA method, although being an acceptable alternative, was limited due to unsatisfactory precision and accuracy. The purpose of the present study was therefore to modify the ELISA method and LUMSDEN(1980) in order to optimize the assay, so previously evaluated by LARSSON that the thyroxine concentration in canine plasma samples could be more accurately and precisely measured. Material and Methods Plasma samples The plasma samples used in the present study was isolated from clinically healthy dogs and from hospitalized dogs of different ages and sexes. Blood samples were collected into sodium heparinate vials (VacutainersB, Becton-Dickinson; USA), and plasma fractions were isolated by centrifugation of the blood samples immediately after collection. The plasma samples were stored at -55°C until analysis. Plasma samples which were subjected to repeated analysis were stored at 4°C for maximally one week or at -55°C for maximally three months.
T4 analysis The T4 analysis was performed with the Enzymun-TestBcBoehrInper Mannhem) ELISA kit, which was developed for the determination T4 in human serum and plasma, and the analysis was assisted by the System ES-33. In order to achievea more sensitive assay analyzer system Enzymun-Test~(BoehnngerMannhelm) which fitted better to the T,-concentrations in dogs, the original assay protocol was modified accordingly: The standard series was changed to represent kit standards a, (2 . a + b)/3, b, c, d (0.0, 15.8, 47.4, 96.2, and 167.0nmol/l of T,), and the amount of T4 conjugate was reduced to 5 0 % compared to the original protocol by adding half the amount of stock solution to the supplied buffer. Furthermore, the two incubation steps were prolonged, so that the immunoreaction durated for 60min and the enzymeassay for 60min, in stead of twice times 30min as suggested by the manufacturer. Elsewhere, the original protocol suggested by the manufacturer was followed. The concentrations were recorded in SI-units, i. e. nmol/l of T,. Specz5city The antibodies in the test kit were reported by the manufacturer to crossreact 100 ?Lo with L- and D-T4, 20 % with tetraiodothyroacetic acid, 3.5 Yo and 2.9 Yo with L-T, and D-T,, respectively, and below 0.1 % with 3,5-diiodothyrosine and 3-iodotyrosine. Precision The intraassay variation, expressed as the coefficient of variation (CVYO), was determined from the differences between duplicate determinations at low, medium, and high T4 levels in 22, 24, and 8 canine plasma samples, respectively. The interassay variation was assessed by determining the CV% of 14 to 16 determinations of T4 concentration in the same 2 canine plasma pools in different assays. Accuracy Accuracy was evaluated by investigation of the recovery of known amounts of T4,added as small volumes of the kit’s standards to a canine plasma pool with a low T4-content (13.6 nmol/l). Response to exogeneoxs Tbyrotropin-releasing hormone (TRH) administration Blood samples were collected from 8 clinically healthy male Beagle dogs just before and 3 hours per dog. after intravenous administration of a total dose of 100 pg TRH (Protirelin@(Rmhe))
Adaptation of a Commercially Available Enzyme Linked Immunosorbent Assay
743
The dogs were fasted for 15 hours prior to and during the study. Water was offered ad libitum at all times.
Detection limit The detection limit, i. e. the least detectable concentration (LDC) defined as the smallest single result which, with a stated probability (95 Yo), can be distinguished from the zero standard, was calculated as the concentration corresponding to the absorbancy of the maximum binding minus twice the standard deviation (SD); SD was obtained from an estimate of the pooled variance of 16 replicate absorbance determinations. Statistical methods LDC, intra- and interassay variation (Intra- and interassay CV%) were calculated using routine descriptive statistical procedures (Box et al., 1978). Investigation of recovery of added T4a-as accomplished by regression analysis in linear and in logarithmic transformation. The latter was introduced in order to achieve homogeneity of the variance (WEISBERG, 1985). Lack-of-fit (LOF) tests were performed to investigate whether the data fitted the straight-line regression model appropriately. Briefly, the total error variance was divided into an estimate of the pure error variance according to repeated determinations plus the error variance arosen from lack of fitting to the model. In order to test the model fitting, the two types of error variance was compared by an analysis of variance as described by Box et al. (1978).
Results Precision The intra- and interassay variation are presented in Table 1. The intraassay CV% was found to be 6.3 %, 4.4 %, and 4.4 %, respectively for the difference between 22, 24, and 8 duplicate determinations of plasma samples with a low, medium or high T4 concentration, while the interassay CV% for two canine plasma pools was 3.8 YO and 12.5 YO,respectively.
Accuracy In Table 2 the recovery of T4 added t o a pool of low-T, canine plasma shown as an regression analysis o n the data and as a regression analysis following a logarithmic transformation of the data. It can be seen that the recovery was 100 70 (95 %-confidence limit: 95-102), since the slope of the linear regression was not significantly different from 1 (p = 0.46). The logarithmic regression analysis stated the existence of proportionality between added and recovered T4, since the slope was not different from 1 (p = 0.37). Both regression equations fitted a straight line well following nonsignificant LOF-tests (p = 0.42 for both the linear and the logarithmic model).
Table 1. Intra- and interassay variations of determinations of thyroxine in canine plasma samples
Intraassay'):
Interassayb):
N
Mean f SD nmoVl
CV%
22 24 8 16 7
20.5 f 1.3 40.6 f 1.8 60.1 f 2.6 15.9 f 2.0 96.1 f 3.7
6.3 4.4 4.4 12.5 3.8
Based on duplicate determinations of different plasma samples; b) Based on replicate determination of the same plasma-pool; N Number of plasma samples; Mean Mean of thyroxine concentration; SD Standard deviation; CV% Coefficient of variation.
744
JENSEN,
HOIERand POULSEN
Table2 Linear and logarithmic regression analysis of recovery investigations
Recovery: Linear Logarithmic
model
r
p (slope = 1)
p (LOF)
Y=l3.6+0.986.X Y=-0.183+1.053.X
0.998 0.986
0.46 0.37
0.42 0.42
Linear model: (found) = intercept + slope . (added); Logarithmic model: In(Y) = intercept + slope . (In(X)); r: coefficient of correlation; p (slope): probability for the slope of the model = 1; p (LOF): probability for model fitting (Lack-of-fit test).
Detection limit The least detectable concentration was calculated as 3.9 nmol/l (maximum binding 2.75 AU, standard deviation 0.18 AU), thus depicting LDC below the expected T4 concentrationsinplasmafromhealthydogs of9.0-28.0 nmol/l ( R ~ ~ ~ e t 1978), a l . , 16.7-46.3 nmol/l (FELDMAN and NELSON, 1987), 19.3-39.9nmol/l (KANEKO,1989) and below the median T4 concentrations observed in the clinically healthy dogs of the present study (Table 3).
Response to exogeneous TRH administration Determination of theT4 concentration in plasma samples taken from 8 normal dogs just before and 3 hours following intravenous administration of 100 pgT R H gave the result shown in Table 3. The T4concentration prior to T R H injection was 38.0 k 7.4 nmol/l and it increased to 51.8 k 8.5nmol/l 3 hours after the injection of TRH. The average increment in T4 concentration was 13.8 f 9.2 nmol/l. All dogs exhibited an increment in the TT 4 concentration ranging from 1.05 to 1.96 times the pre-treatment concentrations. Discussion The ELISA method evaluated in the present study showed the following assay characteristics: As displayed in Table 1, the intra- and interassay variation of the ELISA method (4 % - 13 %)were acceptable at normal T4concentrations as well as at low T4concentrations. Furthermore, the modified ELISA method appears be more precise, as judged from the intraand interassay variation, compared to the ELISA method described by LARSSONand LUMSDEN (1980), who reported an intraassay variation of 14.6 % and an interassay variation of 21.3 %. The reason for the increased precision of this modified method could
Table 3. Thyroxine (T4)concentrations before and after exogeneous administration of Thyrotropinreleasing hormone (TRH) in 8 clinically healthy dogs
Mean f SD Range
T, at T=O
T, at T = 3
AT4
Ratio
38.0 f 7.4 (25.3 -47.7)
51.8+ 8.5 (39.7 - 65.9)
13.8 f 9.2 ( 2.6 - 29.2)
1.40 f 0.32 (1.05 - 1.96)
T4atT= 0: ConcentrationofT,(nmoI/l) beforeTRHinjection;T,atT= 3: ConcentrationofT4(nmol/l)3 hours post-injection; AT4: Change in T4 concentration following TRH injection; Ratio: Ratio between T4 concentrations, (T = 3) / (T = 0); Mean: Mean of thyroxine concentration; SD: Standard deviation; Range: Lowest and highest value measured.
Adaptation of a Commercially Available Enzyme Linked Immunosorbent Assay
745
be due to the modifications of the protocol: the reduction of the amount of conjugate makes the dose response curve more sensitive. Besides, the introduction of an additional, low standard (15.8 nmol/l) so that the concentration range (dose range) includes two standards in the range of normally occurring concentrations of T4 in dogs makes the determinations more precise. The prolongation of the incubation times relative to the protocol suggested by the manufacturer and used by LARSSON and LUMSDEN (1980), was introduced in order to ensure equilibrium binding conditions. Still, one should take into consideration that the assay characteristics of the commercial kit might have been changed since the analysis was evaluated by LARSSON and LUMSDEN (1980). The accuracy of the ELISA method was investigated by recovery of T4 added to a pool of canine plasma containing a low concentration of T4 (13.6nmol/l). Recovery of the added T4 was outlined in both a linear regression model, Y = 13.6 + 0.986. X and in a logarithmic regression model, Y = -0.1 83 + 1.053 .X (Table 2). As the slope of the logarithmic regression model was not different from 1, it was shown that proportionality existed between added and recovered T4. Further, from linear regression model, the recovery was shown to be 100 % since the slope of the linear regression line was not different from 1. These findings are in contrast to the reported overestimation of the original ELISA method (LARSSON and LUMSDEN, 1980), thus indicating that the ELISA method used in this study offers a more accurate determination of T4 in canine plasma compared to the original ELISA method. In canine hypothyroidism the' basal plasma concentrations of T4 may be below the T4 concentrations normally observed in plasma from clinically healthy dogs. Thus, it is essential that tke ELISA method is able to determine T4concentrations in plasma samples with T4 concentrations lower than normal. The detection limit for the ELISA method used in this study was calculated as 3.9nmol/l, a T4 concentration well below the reported T4 concentrations in healthy dogs of 9.0-28.0 nmol/l (REAPet al., 1978), 16.7-46.3 nmol/l (FELDMAN and NELSON,1987), 19.3-39.9nmol/l (KANEKO, 1989), and also below the T4 concentrations observed before (25.7-47.7nmol/l) and after (39.7-65.9 nmol/l) TRH administration to the 8 dogs in the present study (Table 3). The T4 concentrations found in 8 normal dogs (Table 3) were 38.0 k 7.4 nmol/l prior to TRH administration. This finding is comparable to the findings of FELDMAN and NELSON (1987) and KANEKO (1989) which reported 16.7-46.3 nmol/l and 19.3-39.9nmoll1, respectively, whereas REAPet al. (1978) found a normal range from 9.0 to 28.0nmol/l. The reported reference intervals are subject to variation, and this fact evidences the necessity for local reference values. In order to diagnose the presence of canine hypothyroidism, determination of T4 before and after stimulation of the thyroid gland with either TSH or TRH is frequently used, the reason for this being that a dog with hypothyroidism will not respond sufficiently to the stimulatiuon (FELDMAN and NELSON, 1987; KANEKO,1989). In Denmark, only the TRH stimulation test is used as TSH is not at present commercially available. However, the protocol for the TRH stimulation test has not been fully established, although the determination of the T4concentration prior to and 4 hours after intravenous administration of 200 pg TRH per dog has been proposed (EVINGER et al., 1985; FELDMAN and NELSON, 1987). Moreover, the value of the TRH stimulation test in separating euthyroid and hypothyroid dogs in clinical situations as well as the effects of non-thyroidal illness and and drug administrations on the response to TRH remains to be studied (FELDMAN NELSON, 1987). However, the results of the TRH stimulation test used in this study (i. e. determination of T4concentration prior to and 3 hours after intravenous administration of 100pg TRH per dog) clearly indicated that the expected increase in T4 concentration following TRH stimulation was recognized with the ELISA method, the ratio between post- and pre-TRH concentrations ranging from 1.05 to 1.96. Further, the T4 concentrations obtained by the TRH stimulation were in agreement with the findings reported by LOTHROPet al. (1984) and EVINGER et al. (1985), who, though using other protocols for the TRH stimulation test, also observed an increase of about 1.5 times the basal T4 concentra-
746
JENSEN, HQIERand POULSEN
tion. Evidently, this indicates that the TRH stimulation test used in this study may be used to evaluate thyroid function. In conclusion, using the ELISA method with the modifications described in this study, the thyroxine (T4)concentration in canine plasma samples can be both precisely and accurately determined compared to the original ELISA method. The detection limit of modified ELISA method is well below the T4 concentration observed in plasma from healthy dogs. Furthermore, the T4concentrations determined in plasma from healthy dogs both before and after performing of TRH stimulation test are in agreement with those reported by others. Acknowledgements The skilful technical assistance of Mrs. V.S. ~ S T V E D Tand Mrs. E.THOMSENis gratefully acknowledged. The financial support of the Danish Agricultural and Veterinary Research Council is also gratefully acknowledged.
References BOX, G.E.P., W.G. HUNTER, and J.S. HUNTER,1978: Statistics for Experimenters. John Wiley & Sons, Inc., New York, USA. BULLOCK, L., 1970: Protein-bound iodine determination as a diagnostic aid for canine hypothyroidism. J. Am. Vet. Med. Assoc. 156, 892-899. J. V., R. W. NELSON,and G. D. BOTTOMS,1985: Thyrotropin-releasing hormone stimulaEVINGER, tion testing in healthy dogs. Am. J. Vet. Res. 46, 1323-1325. FELDMAN, E. C., and R. C. NELSON, 1987: Hypothyroidism. Canine Hypothyroidism. In: Canine and Feline Endocrinology and Reproduction (editor: D. PEDERSEN), W. B. Saunders Company Philadelphia, pp. 55-90. KANEKO, J. J., 1989: Thyroid function. In: Clinical Biochemistry of Domestic Animals, 4th ed. Academic Press Inc. San Diego, pp. 630-649. (editor: J. J. KANEKO), LARSSON, M., and J.H. LUMSDEN, 1980: Evaluation of an enzyme linked immunosorbent assay (ELISA) for the determination of plasma thyroxine in dogs. J. Vet. Med. A 27, 9-15. LARSSON, M., 1981: Evaluation of a human TSH radioimmunoassay as a diagnostic test for canine primary hypothyroidism. Acta Vet. Scand. 22, 589-591. and V. A. FADOK,1984: Canine and feline thyroid function assessment LOTHROP,C. D., P. M. TAMAS, with the thyrotropin-releasing hormone response test. Am. J. Vet. Res. 45, 2310-2313. 1978: Thyroxine and triiodothyronine Levels in Ten Species REAP,M., C. CASS,and D. HIGHTOWER, of Animals. Southwestern Vet. 31, 31-34. S., 1985: Applied Linear Regression. 2nd ed. John Wiley & Sons, Inc., New York, USA. WEISBERG,
