A rapid fluorometric enzyme immunoassay for the determination of neonatal TSH from blood spots

Clinica Chimica Acta, 202 (1991) 167-178 0 1991 Elsevier Science Publishers B.V. All rights reserved 0009~8981/91/$03.50

167

CCA 05100

A rapid fluorometric enzyme immunoassay for the determination of neonatal TSH from blood spots Tamara Tuuminen ‘, Airi E.M. Rakkolainen ‘, Maj-Gret I. Welin ‘, Theodor H. Weber 2, Pekka L. Nylander 3 and Kirsti I. Gpyaho ’ ’ Labsystems Oy, Helsinki, 2 Aurora City Hospital, Helsinki and 3 Helsinki Maternity Hospital, Helsinki (Finland)

(Received 13 March 1991; revision received 2 August 1991; accepted 5 August 1991) Key worh: Neonatal screening; Congenital hypothyroidism; Fluorometric enzyme immunoassay

Summary We describe a novel method for the detection of thyrotropin from dried blood spots using a horseradish peroxidase-labelled sandwich enzyme immunoassay with fluorometric detection. The detection limit of the present assay is 1.25 mIU/I with within-run and between-run imprecision being in the range 5.2 to 11.4%. The results of the assay correlate well with two commercial methods: an enzyme immunoassay (r = 0.93) and a time-resolved fluorescence assay (r = 0.90). The blood spot values also show a good correlation (1. = 0.93) with respective values obtained from plasma using a commercial immunoradiometric method. The assay may also be performed calorimetrically with sensitivity similar to the fluorometric assay. However, the latter provides a wider dynamic range with an upper limit of 400 mIU/I while the calorimetric method reaches a plateau at 25 mIU/I. Due to its simplicity and rapid performance (3 h), the fluorometric assay is suitable for the routine screening of congenital hypothyroidism.

Introduction Congenital hypothyroidism (CH) is an endocrine disorder in newborns with an average prevalence of 1:4,000 among different ethnic groups with a higher prevalence in iodine-deficient geographic regions. When left untreated CH leads to severe disturbances in psycho-motor development and mental retardation. Due

Correspondence

to: T. Tuuminen, Labsystems Oy, Puhtitie 8, SF-00880 Helsinki, Finland.

168

to the availability of reliable diagnostic methods and replacement therapy, this disease has been included in screening programs in a majority of countries along with some inborn errors of metabolism, e.g. phenylketonuria and galactosemia. The screening of CH is generally based on the immunochemical determination of thyrotropin (TSH) from newborn blood. A variety of radioimmunoassays [l-3] as well as non-isotopic immunoassays [4-131 for TSH have been published and are used in routine screening laboratories. We have recently reported a microplate modification of the McCaman and Robins fluorometric method for phenylketonuria (PKU) screening [14], which is now used in routine practice. To make the combined screening of CH and PKU convenient and economic we decided to develop a fluorometric microplate assay for neonatal TSH to be run simultaneously with the PKU test using the same microplate fluorometer with a high throughput. To make the fluorometric detection of TSH feasible, we have adapted 3-hydroxyphenylpropionic acid (HPPA), a fluorogenic substrate for horseradish peroxidase (HRP), to microplate enzyme immunoassay [15]. The original serum assay has now been modified for the determination of TSH from dried blood spots: the procedure is a one-step microplate enzyme immunoassay with fluorometric detection. The method is simple, rapid, easily mechanised, and can be run simultaneously with the fluorometric PKU screening test. Materials and methods Specimens

In Finland CH screening is concentrated to one laboratory of the National Health Institute where serum samples taken from the umbilical cord blood are being analyzed [16]. Because of this practice and for ethical reasons we performed part of our clinical evaluation using cord blood samples from Finnish healthy newborns prepared as follows. Cord blood samples from 47 healthy full term neonates were obtained from the Helsinki Maternity Hospital (Helsinki, Finland). Blood from each newborn was collected into tubes containing Sodium heparinate, and spotted onto Schleicher and Schuell 903 filter paper in 50 ~1 drops. The paper was allowed to dry overnight and then stored in a plastic bag with a drying agent at + 4 o C, until being analyzed. Examination of hematocrit was performed from each blood sample. Plasma samples for thyrotropin analysis were obtained by centrifugation at 3,000 X g for 15 min of the heparinized blood. Blood specimens from newborns with raised TSH concentrations (n = 61, and who were candidates for CH, were obtained from the Research Institute of Hereditary Diseases (Minsk, USSR). The specimens were collected onto Schleicher and Schuell 2992 filter paper by a heel prick on the fifth day of life for the routine screening of PKU and CH. These blood specimens were analyzed in the Institute by an immunofluorometric (tr-FIA) neonatal TSH test (Pharmacia Diapro, Turku, Finland).

169

Additionally we received 202 blood samples taken onto Schleicher and Schuell paper 2992 by a heel prick from the All-Union Phenylketonuria Centre (Moscow, USSR). Blood calibrators

TSH blood calibrators were prepared from the blood of a healthy volunteer whose hematocrit was adjusted to 50%. Human TSH (Scripps Laboratories, CA, USA) was then added to the blood preparation to obtain calibrators of 2.5, 5, 10, 25, 50, 100, 200, and 400 mIU/I. Calibrators were applied on to Schleicher and Schuell filter papers 903 and 2992. Reagents

3-p-Hydroxyphenylpropionic acid and Tris[hydroxymethyl]aminomethane (TRW were purchased from Sigma Chemical Co (St. Louis, MO, USA), HRP from Boehringer (Mannheim, FRG), hydrogen peroxide (Perhydrol), dimethylsulphoxide (DMSO), 3,5,3’5’-tetramethylbenzidin (TMB), 2-[4-(2-hydroxyethyl)-l-piperazinyl]-ethanolsulfonic acid (HEPES) as well as all salts for buffers which were of analytical quality were from Merck (Darmstadt, FRG). Human thyrotropin was obtained from Sigma. Monoclonal antibody against the hTSH beta-subunit was purchased from OEM Concepts (Tams River, NJ, USA) and monoclonal antibody against the intact TSH molecule was produced by Labsystems Oy (Helsinki, Finland). Neonatal TSH EIA kit was purchased from Fujirebio (Tokyo, Japan), neonatal immunofluorometric TSH kit from Pharmacia Diapro (Turku, Finland) and serum TSH immunoradiometric (IRMA) kit from Behringwerke (Marburg, FRG). Equipment

Microplate fluorometer (Fluoroskan 21, microplate photometer (Multiskanl, transparent and black microplate strips were manufactured by Labsystems. Standard linear regression was carried out to assess correlation between methods. Procedures Determination of TSH from blood dkks by FEIA and EIA

TSH was determined from paper disks using one-step sandwich enzyme immunoassays with a calorimetric (EIA) and a fluorometric detection (FEIA). Two murine monoclonal antibodies were used. The antibody directed against intact hTSH molecule was conjugated with HRP using the maleimide method of Ishikawa et al. [17] and the other antibody against hTSH beta-subunit was used to coat the microstrip wells. The assays were performed similarly, except that the substrate reaction was optimized separately for each assay: the incubation time in FEIA

170

being longer to achieve the same sensitivity [15]. The procedure of FEIA and EIA is outlined as following. One 5-mm disk was punched out of the filter paper and transferred to the microplate well. 200 ~1 of HRP-conjugate in 20 mmol/l Hepes buffer, pH 7.0, containing 1 g/l bovine serum albumin (BSA), 60 ml/l bovine serum, 150 mmol/l sodium chloride and 0.1% Tween 20 was added to each well and incubated for 2 h on a shaker at room temperature. Paper disks were removed and the wells were washed four times with 10 mmol/l phosphate, containing 150 mmol/l sodium chloride and 0.1% Tween 20, pH 7.4. The fluorogenic substrate reaction was performed by adding 200 ~1 of substrate mixture containing 5 volumes of 45 mmol/l HPPA in 0.1 mol/l Tris buffer, pH 7.8, and one volume of 18 mmol/l hydrogen peroxide [15]. The substrate mixture was incubated in a shaker for 1 h at room temperature and the reaction was stopped with 100 ~1 of 1.5 mol/l glycine buffer, pH 10.3. The fluorescence was measured at 405 nm (excitation at 320 nm). The calorimetric detection was performed by adding to the wells 200 ~1 of 20 mmol/l TMB in DMSO diluted 1: 50 with 1.5 mmol/l hydrogen peroxide in 0.1 mol/l sodium acetate adjusted to pH 6.0 with citric acid (1.55 mmol/l final concentration) and incubating for 30 min on a shaker at room temperature. The reaction was stopped by adding 100 ,ul of 2 mol/l sulphuric acid and the absorbances were measured at 450 nm. Determination of TSH in serum by FEIA and EL4

The procedures for serum TSH determination using fluorogenic mogenic substrates have been described earlier [15].

and chro-

Determination of TSH from blood disks by commercial tests

We used a commercial neonatal TSH EIA kit produced by Fujirebio and a commercial neonatal tr-FIA TSH kit manufactured by Pharmacia Diapro. However, to make results comparable we used our calibrators on Schleicher and Schuell paper 903 for the blood specimens obtained from the Helsinki Maternity Hospital Institute. The specimens obtained from the Centre of Inherited Diseases and from the All-Union Phenylketonuria Centre we assayed using Delfia TSH kit with calibrators provided in the kit. Comparison of calculated and measured plasma TSH values

To assess the accuracy of the FEIA test, we converted the TSH blood values into plasma values by correcting for the hematocrit. The obtained values were compared with the true plasma values measured by own FEIA and EIA, as well as with the values obtained with a commercial IRMA kit (Behringwerke). Inteqerence

by heterophilic antibodies

Cord plasma specimens of 47 neonates were assayed in EIA as described’[l5] with and without the use of bovine serum in the conjugate diluent [lS]. Interference by heterophilic antibodies was indicated by a decrease of obtained TSH values in the presence of bovine serum [181.

171

Interference similarly.

by heterophihc

antibodies in blood spot specimens was determined

Bilimbin interference

Interference by high concentrations of bilirubin in the blood spot assay was tested by assaying TSH calibrators containing 0 and 50 mIU/l with and without 100 ~mol/i and 200 pmol/l of bilirubin.

Recovery study was performed by spiking of blood taken from four adult volunteers with various quantities of thyrotropin. After addition of TSH, the blood was spotted onto the fiber paper and dried as usual. The rest&s are averages of four replicates. Results Technical performance

of the neonatal TSIi FEU

The dynamic range of the TSH FEIA was 2.5 to 400 mIU/i, whereas the dynamic range of TSH EIA reached a plateau already at 25 mIUfl although the sensitivity of the assay was simifar with the FEIA. The comparison of calibration curves of the neonatal FEIA vs. EIA is shown in Fig. 1. The detection limit of the assay was calculated from the mean fluorescence of the zero standard pIus 3 SD. This calculation gave a detection limit of 1.25 mIU/I. Fluorescence

OD (450 nm)

units 10000

10

1000

1

100

0.1

10

0.0 1 1

100

10 TSN ( mlU/LI

Fig. 1. Catibration curves of TSH FEIA vs. EIA.

1000

172

TABLE I Intra- and inter-assay imprecision of the TSH FEIA Sample

Mean TSH mIU/I

No. of replicates

No. 41 a No. 47 ’ T-r25 h MWlOO h

5.5 13.0 10.0 24.7 100.3

15 16 13 16 16

Sample

Mean TSH mIU/I ’

No. of assays

CV%

6 6 6 6

11.4 5.7 5.2 5.2

Intra-assay imprecision No. 20 ’

Inter-assay imprecision No. 35 a lT5h ‘IT25 h TTlOO h

4.9 6.6 24.9 95.0

CV%

9.2 7.3 7.7 8.2 6.8

’ Newborn cord blood. ’ Spiked blood from adult volunteers. ’ Mean of duplicates.

The within-run and between-run imprecision were studied using cord blood specimens from newborns (when possible) and blood samples from volunteers spiked with different quantities of thyrotropin. Data are presented in Table I. Results of the recovery study are presented in Table II.

TABLE II Analytical recovery of the TSH FEIA Endogenous

2.3

1.5

0.7

1.7

Recovery (%l

Thyrotropin (mIU/I) Added

Measured

5 25 100 5 25 100 5 25 100 5 25 100

7 24.5 98 7 23 100 5.5 24 95 6.9 25 87

93 90 96 107 87 98.5 96 93 94 102 94 85.5 Mean: 94.6

0

0

2

4

2

6

4

8 10 TSX (mlU/L) EIA

6 8 TSH (mlU/L) FIA

12

10

14

12

16

14

Fig. 2. A. Correlation of TSH FEIA vs. EIA by Fujirebio. Regression equation y = 0.91 x + 0.35, where x = TSH values in EIA and y = corres~nding values in FEIA, r = 0.93, n = 47. B. Correlation of TSH FEIA vs. tr-FIA by Diapro. Regression equation y = 1.16~ - 0.45, where x = TSH values in tr-FIA and y = corresponding values in FEIA, r = 0.90, n = 47. C. Correfation of calculated TSH plasma values using hematocrit vs. measured plasma TSH values in IRMA test by Behring. Regression equation y = 1.11.x -0.77, where x = TSH values in IRMA test and y = corresponding calculated TSH values, r = 0.93, n = 42.

174 TSH (mlU/L) calculated plasma values

40 C

I

0

0

I 5

I

10

I

i

I 15 20 TSH (mlU/L) IRMA

t 1 25

I 30

35

Fig. 2 (continued).

Clinical performance

of the neonatal TSH FEIA

Correlation with three commercial assays was studied using dried cord blood samples of 47 neonates (see ‘Materials and Methods’). The results are presented in Fig. 2. Comparison of the present method with tr-FIA (Pharmacia Diapro) was also performed on 202 blood spot samples taken by a heel prick. The results showed good correlation giving regression equation: y = 0.94~ + 3.02 (r = 0.89), where y were values obtained by our method and x were corresponding values by tr-FIA. Comparison of TSH FEIA with tr-FIA performed on samples suspected for CH is shown in the Table III.

TABLE III Comparison of the results obtained by FEIA and tr-FIA on six blood spot specimens suspected for congenital hypothyroidism Neonatal sample

Ger. Bes.

Dro. Zel. Shi. Zyb.

TSH mIU/l FEIA

tr-FIA

>400 165 81 140 240 52

367 191 89 141 234 64

175 TABLE IV Relative increase of TSH values by heterophilic antibodies Increase in TSH (%o)

% of sera n = 42

% of blood disks n = 47

< 30 31-100 101-500 > 500

47 30 14 9

96 4 _

Interferences

We observed distinct interference (> 30% increase of TSH value) by heterophilic antibodies in 23 out of 43 neonatal plasmas but only in two blood disks (Table IV). Interestingly, the two blood disk specimens showing interference both also showed very high (more than 5-fold) false increase of the plasma TSH values. There was no interference by bilirubin at concentrations of 100 kmol/l and 200 pmol/l. Discussion Since 1975, when the first description of the measurement of TSH from dried blood spots appeared [1,2], a variety of non-isotopic techniques [4-131 as alternatives to radioimmunoassays have been described, some of them possessing a high sensitivity [13]. However, these assays have been difficult to automate because of the tube format [11,13] and thus have not been feasible for the mass screening in laboratories with a throughput of up to 1,000 samples daily. The present semi-automatic assay was developed to overcome these difficulties by using microplate formate fluorometer. The assay shows good within-run and between-run reproducibilities: CVs ranged from 5.2 to 11.4% in the concentration range from 4.9 to 100.3 mIU/l (Table I). The assay showed good correlation with the commonly used commercial neonatal TSH tests, the enzyme immunoassay of Fujirebio (Fig. 2A) and tr-FIA of Diapro (Fig. 2B, Table III>. In our study, we paid special attention to the most frequent source of interference in two-site immunoassays, the presence of heterophilic antibodies [8,18]. These heterophilic antibodies arise in humans due to the immunization by heterologous proteins, e.g. from milk or beef meat products. Heterophilic antibodies bind unspecifically to monoclonal antibodies of murine origin used in the assay interfering with them and thus causing false results. It has been found that heterophilic antibodies being of IgG-class readily penetrate through placenta and are detectable also in neonates. Indeed, we observed interference (Table IV) in 23 out of 43 plasma specimens and in two out of 47 blood disk specimens. The prevalence of heterophilic antibodies in our limited cohort of neonates is essentially comparable

176

to the prevalence in adult population [18] although the increases appear to be milder than in adults. The fact that blood disk specimens only rarely show interference (Table IV) is explained by the relatively small volume of the blood spot sample, which is approximately equal to 10 ~1 of blood compared to 100 ~1 of serum used in the serum assay. The interference can be overcome by dilution of the specimen or by addition of neutralizing IgG (eg. serum) to the assay [18] which was used in the present study. When both substrate reactions are optimized, essentially equal sensitivities are achieved, but fluorometric detection facilitates obtaining a much wider dynamic range (in our study from 2.5 to 400 mIU/l) compared to the calorimetric assay which reaches a plateau at 25 mIU/l (Fig. 1). Ishikawa et al. [13] using 4-methylumbelliferryl-beta-D-galactoside as a substrate for Fab’-beta-D-galactosidase conjugate, achieved an extremely wide dynamic range from 0.037 to 370 mIU/l, thus enabling the diagnosis of both primary hypothyroidism with elevated TSH values as well as of secondary and tertiary hypothyroidism of neonates. Woodhead et al. reported a immunochemiluminometric assay of thyrotropin from blood spots [ll] using acridinium ester-labelled monoclonal antibody. The dynamic range of this assay being from 6.25 mIU/l to 200 mIU/l, was comparable with that of the described FEIA. The incidence of pituitary-hypothalamic hypothyroidism varies, depending on the population studied, from 1: 69,000 [19] to 1: 1,300,000 [20] and due to its low prevalence, screening of secondary-tertiary hypothyroidism is not considered to be cost-effective [21]. As pathologic TSH concentrations in primary CH are far beyond 20 mIU/l, we conclude that detection of very low TSH values in neonatal mass screening on primary CH is not of primary importance, whereas rapid performance, reliability and low cost of the assay are prerogatives for implementation of the test into a mass screening program. Our experience shows that the present method is suitable for this purpose. Its validity will be further documented in a large scale international multi-centre study. Acknowledgements

We are grateful to Dr. G.L. Zukerman from the Research Institute of Hereditary Diseases (Minsk, USSR) and to Dr. A.D. Baikov from the All-Union Phenylketonuria Centre (Moscow, USSR) for their valuable samples. References 1 Irie M, Enomo K. Measurement of thyroid stimulating hormone in dried blood spot. Lancet. 1975;20:1233-1234. 2 Illig R, Torresani T, Sobradillo B. Early detection of neonatal hypothyroidism by serial TSH determination in dried blood. Helv. Paediatr. Acta 1977;32:289-297. 3 Giga M, De Piro Cl, Mansbach L. Development of solid phase neonatal TSH immunoradiometric assay. Clin Chem 1990;36,Nr.6:0594. 4 Torresani TE, Qui-Qing, Illig R. Thyrotropin enzymeimmunoassay in dried blood spots: a spectrophotometric method for neonatal thyroid screening. J Clin Chem Biochem 1986;24:199-203.

5 Torresani TE, Scherz R. Thyroid screening of neonates without use of radioactivity: evaluation of time-resolved fluoroimmunoassay of thyrotropin. Clin Chem 1986;32:1013-1016. 6 Volbracht L, Beyer P, Weber B, Helge H. TSH measurements from blood spots on filter paper by enzyme immunoassay (EIA) alternative to RIA as screening procedure for neonatal hypothyroidism. Advances in neonatal screening Amsterdam: Elsevier Science Publishers; 1987;153-155. 7 Arends J, Norgaard-Pedersen B. Immunofluorometry of thyrotropin from whole blood spots on filter paper, to screen for congenital hypothyroidism. Clin Chem 1986;32:1854-1856. 8 Norgaard-Pedersen B, Nielsen J. Screening for congenital hypothyroidism by time-resolved immunometric TSH analyses. A 14 month experience with 71258 paper disc samples. Amsterdam: Elsevier Science Publishers, Advances in neonatal screening, 1987;129-131. 9 Suzuki N, Yokota M, Shirane H. Enzyme immunoassay of TSH for neonatal screening. Amsterdam: Elsevier Science Publishers. Advances in neonatal screening 1987;165-166. 10 Miyai K, Hata N. Endo Y, Iijima Y, Amino N. lshibashi K et al. Semiautomated enzyme immunoasay for neonatal hypothyroid screening. Elsevier Science Publishers. Advances in neonatal screening 1987;123-124. 11 Woodhead JS, Siddle K, Weeks J. Immunochemiluminometric assay of thyrotropin in filter paper blood spots. Amsterdam: Elsevier Science Publishers, Advances in neonatal screening, 1987;149-152. 12 Naruse H, Suzuki, E, Kumada J. Non-radioisotopic methods for neonatal hypothyroid screening. Amsterdam: Elsevier Science Publishers. Advances in neonatal screening, 1987;115-119. 13 Ishikawa E, Hashida S, Umehashi H, et al. Highly sensitive sandwich enzyme immunoassay of human thyroid stimulating hormone (hTSH) in dried blood on filter paper discs for mass screening of neonatal hypothyroidism. Amsterdam: Elsevier Science Publishers. Advances in neonatal screening 1987;121-122. 14 Gerasimova NS, Steklova IV, Tuuminen T. Fluorometric method for phenylalanine microplate assay adapted for phenylketonuria screening, Clin Chem 1989;35:2112-2115. 15 Tuuminen T, Palomiki P, Rakkolainen A, Welin M-G, Weber T, tipyaho K. 3-p-Hydroxyphenylpropionic acid - a sensitive fluorogenic substrate for automated fluorometric enzyme immunoassays. J Immunoassay 1991;12:29-46. 16 Virtanen M, Perheentupa J, Maenpla J, Pitkanen L, Pikkarainen J. Finnish national screening for hypothyroidism. Eur J Pediatr 1984;143:2-5. 17 Ishikawa E, Imagawa M, Hashida S, Yoshitake S, Hamaguchi Y. Ueno T. Enzyme-labeling of antibodies and thier fragments for enzyme immunoassays and immunochemichal staining. J Immunoassay 1983;4:209-327. 18 Weber TH, Kapyaho KI, Tanner P. Endogenous interference in immunoassays in clinical chemistry. A review. Scan J Clin Lab Invest 1990: Suppl. 201:77-82. 19 Schroenberg D. Klett M. Screening for congenital hypothyroidism in the Federal Republic of Germany. Past, present and future. Amsterdam: Elsevier Science Publishers, Advances in neonatal screening 1987;25-26. 20 Wilken B, Brown A, Raby J, Connely J, Francis I, McFarlane J, Bowling F. Newborn screening for metabolic disorders in Australia and New Zealand: results for 1983. Med J Australia 1985;143:159,160. 21 Fisher DA. Effectiveness of newborn screening programs for congenital hypothyroidism: prevalence of missed cases. Ped Ado1 Endocrinol 1987;34:881-890.