Highly Sensitive Assays of Autoantibodies to Thyroglobulin and to Thyroid Peroxidase

Transcription

1 CLIN. CHEM.35/9, (1989) Highly Sensitive Assays of Autoantibodies to Thyroglobulin and to Thyroid Peroxidase K. Beever, J. Bradbury,. PhillIps,2S. M. McLachlan,2C. Pegg,3 A. Goral,4 W. Overbeck,4 G. Felfel,5 and B. Rees Smith2 These highly sensitive assays are based on the interaction between thyroid autoantibodies and 125l-labeled autoantigens. Serum samples are incubated with labeled thyroid peroxidase (TPO) or thyroglobulin (Tg) to allow the formation of antibody-labeled antigen complexes. The complexes are then precipitated by addition of solid-phase Protein A. In the presence of high concentrations of TPO antibody or Tg antibody, more than 5% of the respective labeled antigen was precipitated, whereas only 1-2% was precipitated in the absence of autoantibody. Interassay CVs were 3.2% and 5.7%, respectively, for the anti-tpo and anti-tg assays. There was no cross-reactivity between Tg antibody and TPO antibody. Results correlated highly significantly with results from other assay systems based on antigen-coated cells or plastic supports, but the assays described here were considerably more sensitive. Scatchard analysis of the assay data provided information on the affinity and serum concentration of TPO autoantibodies (ka - 1o LJmol and concentrations up to 1 g/l) and Tg autoantibodies (k5-4 x 1#{176} Llmol and concentrations up to 1 gil). Overall, these assays provide a sensitive, precise, and convenient system for measunng and investigating the properties of thyroid autoantibodies. immunoradiomefricas- Additional Keyphrases: thyroidstatus say Currently, several different techniques are used to detect and (or) quantify autoantibodies to thyroglobulin (Tg) and to thyroid peroxidase (TPO; microsomal antigen), including systems based on antigen-coated cells or particles (1), antigen-coated ELISA plates (2), antigen-coated tubes (3), or monoclonal antibody-coated tubes (4).#{176} Although these assays systems can give useful clinical information, they all depend on indirect measurement of the interaction between autoantibody and autoantigen, which can result in limitations of sensitivity, precision, and ease of handling. To overcome these limitations, we have developed an assay system based on the direct interaction between 1251-labeled autoantigens and autoantibodies. Materials Serum Samples and Methods Serum samples were obtained from patients visiting the thyroid and rheumatology clinics and from healthy labora- 1RSR Ltd., Avenue Park, Pentwyn, Cardiff, CF2 7HE, U.K. 2Endocrine Immunology Unit, 7th Floor Medicine, University of Wales College of Medicine, Cardiff, CF44XN U.K. 3Department of Surgery, University of Nottingham, Nottingham, U.K. 4Chirurgische Klinik, Stadt Krankenhaus, Kaiserslautern, F.R.G. 5Chirurgische 6Nonstandard Uni-Klinik, Homburg/Saar, F.R.G. abbreviations: Tg, thyroglobulin; TPO, thyroid peroxidase/microsomal antigen; ELISA, enzyme-linked immunosorbent assay; and TRAb, thyrotropin receptor antibodies. Received April 18, 1989; acceptedjune 28, tory and hospital staff Patients with Graves disease were diagnosed on the basis of clinical symptoms; the presence of TRAb, as measured by the method of Southgate et al. (5); and biochemical confirmation of a history of hyperthyroidism. All patients with Hashimoto s disease had thyroid failure in association with thyroid microsomal antibodies (measured by Wellcome tanned red cell test) and were being treated with thyroxin. Patients with multinodular goiter and rheumatoid arthritis were assayed clinically and multinodular goiter was confirmed by thyroid scan. All sera were stored in aliquots at below -2 #{176}C before assay. Thyroid Autoantibody Reference Preparations Anti-thyroid microsome serum (66/387) and MRC antithyroglobulin antibody research standard A (65/93) were obtained from the National Institute for Biological Standards and Control, Hertfordshire, U.K. Each vial, designated as containing 3 units of antibody, was reconstituted with normal serum (autoantibody negative), diluted 2-fold in serum diluent (per liter, 15 mmol of NaC1, 1 mmol of Tris HC1, ph 7.5,5 g of bovine serum albumin, and 1 ml of Tween 2). Aliquots were then further 2-fold diluted in normal serum to give a wide range of antibody concentrations. Sera containing high concentrations of autoantibodies to Tg or to TPO were pooled and stored at -4#{176}C for use in preparing an in-house reference material. This preparation was then calibrated against the Tg antibody and TPO antibody reference preparations 66/387 and 65/93. Procedures Preparation of thyroid autoantigens. Human thyroid tissue was obtained from patients undergoing partial thyroidectomy for Graves disease, and a microsomal fraction was prepared by differential centrifugation of tissue homogenates (2). The microsomes were then solubilized in sodium deoxycholate as described by Kajita et al. (6). Thyroid peroxidase in the deoxycholate-solubilized microsomes was then purified by affinity chromatography, as described by Czarnocka et al. (7) and Ohtaki et al. (8) and stored in aliquots at -7 #{176}C. The purified preparations had a specific activity of 2-3 arbitrary units of TPO activity per milligram of protein [TPO activity was measured by the guaicol method of Hosoya and Morrison (9) and protein concentration by the method of Bradford (1)], and analysis by polyacrylamide gel electrophoresis indicated that the major component migrated as a double band of Mr 1, as described previously (6). Thyroglobulin contamination of the purified TPO preparations was.2% as judged by Tg radioimmunoassay (11). Thyroglobulin was purified from the supernates obtained after sedimentation of thyroid microsomes by precipitation between 1.52 and 1.76 molil ammonium sulfate (12) followed by gel filtration on Sephacryl S-4 (obtained from Pharmacia, U.K. Ltd., Milton Keynes, U.K.). Aliquots were stored at -7 #{176}C. Labeling of TPO and Tg with 125J TPO was labeled to a specific activity of 1 Ci per gram of protein by using the CLINICALCHEMISTRY,Vol. 35, No. 9,

2 lodogen method (13) and Tg to the same specific activity by using the Chlorainine-T method (14). Labeled materials were purified by gel filtration on Sephacryl S-3 (TPO) or S-4 (Tg) and stored in aliquots at -7 #{176}C. Interaction of thyroid autoantibodies in serum with 125j.. labeled TPO and Tg. After diluting serum samples 2-fold in serum diluent, we incubated 5-pL aliquots of diluted sera in triplicate with 5 1.iL of lsilabeled TPO or Tg (each at 4 counts/mm for various times and at various temperatures). Solid-phase Protein A (5 j.tl) in the form of Staphylococcus aureus cells (1 g/l suspension of Pansorbin from Calbiochem, La Jolla, CA) was then added and incubated for 6 mm at room temperature. We then added 1 ml of serum diluent, centrifuged the tubes (15 x g for 3 mm at 4 #{176}C), and counted the 1251 radioactivity in the pellets. ELISA assays for Tg antibodies and TPO antibodies. We used the method of Schardt et al. (2). Briefly, 2-folddiluted serum samples were added to ELISA plates coated with human Tg or thyroid microsomes. Antibody binding was assessed by a peroxidase-conjugated anti-igg/o-phenylenediamine system. Coated-tube immunoradiometric assay for Tg antibodies and TPO antibodies. In this procedure (3), 2-fold-diluted serum samples were added to antigen-coated tubes and antibody binding was assessed by using msl.labeled Protein A. 2. SC I- in N 2. C- in N..5 1oo 6. 2G 6, a) Tg antibody (units ml b) Results Effect of Time and Temperature on Antigen/Antibody Interactions Anti-Tg antibodies. The effect of time and temperature on the interaction between labeled Tg and Tg autoantibodies is shown in Figure 1 and Table 1. Antibody binding increased with both time and temperature, but the radioactivity bound in the presence of the zero standard was essentially unchanged. Further studies (not shown) indicated that binding was maximal between 2 and 48 h at 37 #{176}C. Anti-TPO antibodies. The binding of anti-tpo antibodies to labeled TPO showed considerably less time and temperature dependence than the Tg antigen/antibody system. Binding was essentially unchanged between 1 and 2 h at 37 #{176}C except for a small increase in binding in the presence of the zero standard. Similarly, increasing the temperature from 4 to 2#{176}C increased binding only slightly after 4 h. Specificity of Antigen/Antibody Interactions Sera from patients with Hashimoto s disease, which contained both anti-tg antibody and anti-tpo antibody, were diluted in the presence of cold Tg, and the cold Tg had no effect on labeled TPO binding (Figure 2), but markedly reduced Tg tracer binding (Figure 2 and Table 2). Similarly, unlabeled TPO inhibited the binding of labeled TPO but had no effect on the binding of labeled Tg (Table 2). Nonspecific binding. The effects of unlabeled Tg or TPO on labeled-antigen binding in the presence and absence of antibody are also shown in Table 2 and Figure 2. Radioactivity bound in the presence of the zero standard was not influenced by addition of unlabeled antigen, indicating that this binding was nonspecific and did not represent the presence of low quantities of autoantibody in the pool of normal serum used to prepare the zero standard. The TPO antibody (units mi1) Fig. 1. Time courseof(a) the Tg to anti-tgabinteractionsat different antibody concentrations (at 37#{176}C), (b) the TPO to anti-tpoab interactions at different antibody concentrations (at 2#{176}C) #{149} = 1 h; = 4 h;#{149} = 2 h aftermixing antigenandantibody Table 1. Effect of Temperature on the Tg to Antl-TgAb and TPO to Antl-TPOAb Antigen-AntIbody Interaction % I-Iabe4ed antigen bound after 4 ftat: 4#{176}C Anfi-TgAbconcn,unifs/mL Anti-TPOAbconcn, units/ml #{176}C #{176}C quantities of labeled antigen bound in the presence of antibody could be reduced to values similar to those obtained with the zero standard binding by adding unlabeled antigen (Table 2 and Figure 2). Estimation of Binding Affinities Scatchard analysis (15) of the Tg-anti-Tg antibody interaction and the TPO-anti-TPO antibody interaction is 195 CLINICALCHEMISTRY, Vol. 35, No. 9, 1989

3 C. C a.. I-. I- C Ie N I-. a 1 o -a-- 1:8 1:2 Hashimoto serum dilution -o 1:5 Fig. 2. Effect of unlabeled Tg on binding of labeled Tg or labeled TPO to their respective autoantibodies in different dilutions of serum from a patientwith Hashimoto sdisease #{149}, only;, Tgplus cold Tg, 2 mg/l;s, 251.Tgplus cold 1g. 1 mg/l;, 1251-TPOplus cold Tg,, 2, and 1 mg/i. (all points conjoint) shown in Figure 3. The plots were approximately linear. Estimates of the association constants and serum antibody concentrations, derived from analysis of different individual sera, are shown in Table 3. The anti-tg antibody-tg interactions were of high affinity and all the samples studied had similar association constants: range, 2.8 x 1 #{176} L/mol to 4.9 x 1#{176} L/mol. Antibody concentrations, estimated from the Tg-binding capacity, ranged from.3 to.72 g/l. Anti-TPO antibodies had lower association constants than anti-tgab: range, 8.2 x 18 L/mol to 1.1 x 1#{176} Lfmol (Table 3 and Figure 3). Anti-TPO antibody concentrations ranged from.5 to 1.4 g/l. Routine Procedure for Assay of Anti-TgAb and Anti-TPOAb The interaction between labeled thyroid antigen and antibody was clearly suitable for measuring anti-tpoab and anti-tgab in serum and we used the following routine procedure: Incubate for 1 h at room temperature 5 tl of serum (in triplicate) diluted 2-fold and 5 p.l of the labeled Tg or TPO, then add 5 pl of solid-phase Protein A and incubate for a further 1 h at room temperature. Add 1 ml of serum diluent and centrifuge at 15 x g at 4#{176}C for 3 mm, to separate bound and free labeled antigen. Under these conditions, intra-assay variation (CV) of the anti-tpoab assay was 2.% at a mean value of 3.9 units/ ml (n = 5), and interassay CV was 3.2% (mean value = 3.8 units/ml, n = 5). In the anti-tgab assay, intra-assay CV was 3.2% at a mean value of 1 unit/ml (n = 5), and interassay CV was 5.7% (mean value, 1.2 units/ml, n = 5). Table 2. Effect of Unlabe led Antigen on Bind Ing of Labeied Antigen % of 1251-labeledTPO boundinthe presence of % of 2l.label.d Tg bound Inthe presenceof Unlabeled TPO and Tg AntI-TPOAb concn, unlabeled TPO at Antl-TgAb concn, No unlabeled Unlabeled TPO, Unlabeled Tg, unhs/ml mg/i. 2 mg/i. units/mi. TPO or Tg 2 mg/i. 2 mg/l 2.8 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ±.4 Results are mean± SD of n = 5 determinationseach p Bound Free.8#{149} Bound Free.6.4 a Tg bound (ng) TPO bound (ng) Fig. 3. Scatchard analysis of the (left) Tg to anti-tgab interaction and (right) TPO to anti-tpoab interactionin dilutedserum samplesfrom different patients Different symbols represent different patient s sera. See text for full experimental details andtable 3 for valuesofassociation constants and serum antibody concentrationsestimated from the data CLINICAL CHEMISTRY, Vol. 35, No. 9,

4 Table 3. Estimates of Antibody AffinitIes and Concentrations in Serum Serum Anti-TPOantibody k L/mol Antibody concn, g/l 1.1 x i x x x io x i x Anti-Tg antibody 3.9 x x x x assay was set at 2 SD above the zero value, or -=.2 unitlml. Anti-TPOAb. Labeled TPO binding was 1.88 ±.4% (mean ± SD; n = 1) in the presence of the zero standard. As with the anti-tgab assay, this value was not influenced by addition of unlabeled TPO (Table 2), indicating that it represented nonspecific binding and not the presence of TPO antibodies. The limit of detection, 2 SD above the zero standard value, was -.3 unitiml. Effect of Serum from Various Patients Groups Anti-TPOAb. Figure 4 shows the anti-tpoab activity determined in different groups of patients. Two of 2 healthy laboratory and hospital personnel showed anti- TPOAb activity >.3 unitiml, one of these samples (5 unitlml) also gave a positive antibody result by tanned red cell test (titre 1:16). Among the patients, 78 of 9 (87%) of Graves sera, 12 of 23 (52%) rheumatoid arthritis sera, Limit of Assay Sensitivity and 16 of 16 Hashimoto sera contained TPO antibody >.3 Anti-TgAb assay. In the presence of the zero standard, unitlml. In addition, of patients with multinodular goiter, labeled Tg binding was 1.95 ±.2% (mean ± SD; n = 1). four of seven showed detectable amounts of anti-tpoab in As mentioned previously, this value was not influenced by serum (.8, 4.2, 4.8, and 24 units/ml). the presence of unlabeled Tg (Table 2), indicating that this Anti-TgAb. Results for Tg antibodies are also shown in binding represented nonspecific binding and not the pres- Figure 4. Anti-TgAb was <.2 unitlml in all 2 laboratory ence of anti-tgab. The limit of detection in the anti-tgab and hospital personnel studied. Among the patients, 46 of S... >1) : 1 19 i :. 8 S. S.:;: S... I. 5 #{163}. 1 : 1 S. S. #{149}. #{149} S S S..3 #{149} S :.2 : S U-- -UI.m u - Healthy Graves Hashimo(o Rheumatoid Healthy Graves Hashimoto Rheumatoic normal (n = 9) (n=16) arthritis normal (n =69) (n = 13) arthritis donors (TRAb t ye) (n=23) donors (TRAb + ye) (n =23) (n=2) In 2l)) Fig. 4. Concentrations of (left) Anti-TPOAb and (right) anti-tgab measured by assays in various groups of patients MicAb = anti-tpoab;trab +ve Graves = serafrompatientswith treatedoruntreatedgraves disease, which were positivefor thyrotropin receptorantibodies 1952 CLINICAL CHEMISTRY, Vol.35, No. 9, 1989

5 69 (67%) of Graves sera, 13 of 13 Hashimoto sera, and 8 of 23 (35%) rheumatoid arthritis sera contained anti-tgab >.2 unitiml. Of the seven patients with multinodular goiter, two showed detectable anti-tgab (1.2 and 12 units/ ml). In a more extensive study of normal healthy subjects, in which we analyzed serum samples from a blood donor clinic, anti-tgab concentrations.2 unitlml were detectable in 18 of 122 (14.8%) of female donors and anti-tpoab concentrations.3 unitlml in 19 of 121 (15.6%). Comparison u 1:ZS,6[ s I i:ieoo[ l:4e[ 31:512 1:I[ u[ #{149} 1:256 1: :32 1:16 1:5 1:4 of the New Assays with Other Assay Methods Tanned red cell test. Comparison of the new assay for anti-tpoab with the tanned red cell test is shown in Figure 5. All sera positive by the tanned red cell test for thyroid microsomal antibody (titre 1:1) contained detectable antibody by the new assay (.3 unitlml). However, the new assay was considerably more sensitive, with several negative sera in the tanned red cell test showing positive values in the new assay. Overall, the two methods showed a highly significant correlation (r =.69; P <.1; n = 48). Similar results were obtained in a comparison of the new assay and the tanned red cell test for anti-tgab (Figure 5). The two methods were highly significantly correlated (r =.73; P <.1; n = 38), but the new assay was more sensitive. Analysis of the standard sera by tanned red cell test indicated that its lower limit of detection for both anti- TgAb and anti-tpoab was equivalent to 5-1 units/ml. o G,aves (n.21) #{149} Hasbin.oto (n. (6) #{149} RA(n.l1) Mlaoso.n4..stlbody dbod assay {odts n.l) (n4 o G,aves #{149} Hashimoto (,11 #{149} RA(n.6) oo >1(1 Thytoglobulin antibody di,ect assay (Units Fig.5. Comparisonofthe newassaysfor(top) anti-tpoab (microsomalantibody) and (bottom) anti-tgabwiththe respective tanned red cell (TRC) tests Only sera positive in the newassaysare shown.the relationship between assays for anti-tpo antibodiesgave an rvalue of.69 (P <.1) and that between antl-tgantibodiesan r value of.73 (P <.1). RA, rheumatoid arthritis ELISA assays. Anti-TgAb and anti-tpoab measured by ELISA showed a highly significant correlation with values obtained by the new assays (r =.99, P <. 1, n = 39 for anti-tgab, and r =.65, P <.1, n = 48 for anti-tpoab). Again, the lower detection limits of the ELISA5 were equivalent to 5-1 units/ml. Assays based on 251-labeled Protein A. These assays had a similar sensitivity to the ELISA8 (lower detection limit, 5-1 units/ml), and correlations similar to those of the ELISAS were obtained between the new assays and this system (r =.91, P <.1, n = 39 for anti-tgab, and r =.78, P <.1, n = 48 for anti-tpoab). Variations in Standard Assay Procedure Assay time could be reduced by adding diluted serum, tracer, and Protein A suspension together, incubating for 3 mm at room temperature, then adding 1 ml of diluent and centrifuging. This procedure, suitable for a rapid screening assay, had detection limits of 1 unitiml for anti-tpoab and 3 units/ml for anti-tgab. The sensitivity of the anti-tgab assay could be increased by increasing incubation times and (or) temperature (Figure 1 and Table 1). The sensitivity of the anti-tpoab assay, however, was essentially unchanged by increasing incubation times and (or) temperatures. Discussion The interaction between labeled Tg or TPO and anti- TgAb or anti-tpoab could be demonstrated readily by using S. aureus cells to separate antibody-bound tracer. Tg and TPO binding showed differences in time and #{149} temperature dependence as well as differences in associa- #{149} tion constant. In particular, for the interaction between TPO and anti-tpoab, binding was maximal after 1 h at 4, 2, or 37#{176}C. In contrast, greatest binding in the interaction #{149},,. between Tg and anti-tgab occurred after 24 h at 37#{176}C; #{149} binding markedly increased with temperature over the range 4-37 #{176}C. The difference in binding kinetics could be due to a much slower dissociation rate for the reaction between Tg and anti-tgab compared with the TPO and anti-tpoab reaction. Scatchard analysis of the anti-tgab interaction (Figure 3) gave approximately linear plots, although there was an 1 indication of some curvature at low bound/free values. This suggested that most of the labeled Tg was interacting with antibodies of a relatively restricted range of affinities. We #{149}.1 estimated the association constants of the interaction between Tg and anti-tgab from linearized plots (Figure 3), which indicated that antibodies with similar high affinities were present in the different sera studied. However, the concentration of anti-tgab in different sera, calculated from the Tg-bound (y-axis) intercept (Figure 3), showed marked variations between different patients, ranging from.3 to.7 g/l. The estimations of anti-tgab affinity and concentration described above were based on the assumption of a univalent interaction between Tg and anti-tgab. In recent observations in our laboratory (16, 17), a human monoclenal thyroglobulin autoantibody showed the same high affinity by Scatchard analysis as the donors serum antibody did. Furthermore, the concentration of the monoclonal antibody derived from the Scatchard plot was in reasonable agreement with the observed IgG concentration. Scatchard analysis of the interaction between TPO and anti-tpoab (Figure 3) also gave approximately linear CLINICAL CHEMISTRY, Vol. 35, No. 9,

6 plots, suggesting that, as in the case of Tg, most of the labeled TPO was interacting with antibodies of a relatively restricted range of affinities. Estimation of the association constants and antibody concentrations, with the same procedure and assumptions as for interaction between Tg and anti.tgab, indicated that anti-tpoab affinities were similar in different sera; however, the values were considerably lower than those obtained for the interaction between Tg and anti-tgab. Anti-TPOAb concentrations showed marked variations between different patients (range.5 to 1.4 g/l), but the range of values was similar to those for serum anti-tgab concentrations. The differences in association constants between anti-tpoab and anti-tgab were consistent with the different binding kinetics shown in Figure 1 and Table 1. In particular, both the high affinity of the Tg system and the long time required to reach maximum binding to anti-tgab could be due to a much slower dissociation rate constant. In contrast, the interaction between TPO and anti-tpoab has the characteristics of rapid forward and reverse rate constants (relatively low affinity and a relatively short time required to reach maximum binding). In some sera, concentrations of anti- TPOAb or anti-tgab were as much as 1 g/l, thus representing 5-1% of the total serum IgG. In these patients, a similarly high proportion of their antibody synthesizing capacity was presumably committed to thyroid autoantibody production. As well as providing information on the antigen-antibody interaction, the direct binding of antibody to labeled antigen provided a convenient system for measuring antibody concentrations in serum. Unlabeled Tg did not interfere in the anti-tpoab assay (Figure 2 and Table 2), emphasizing the high specificity of the assay. Furthermore, 11 TRAb-positive Graves sera were negative in the anti-tpoab assay (Figure 4), indicating that TRAb did not interfere in this system. Similarly, unlabeled TPO did not interfere in the anti- TgAb assay (Table 2). In addition, 2 TRAb-positive Graves sera from the group shown in Figure 4 gave negative values in the anti-tgab assay, indicating that TRAb did not interfere. Furthermore, nine of these anti- TgAb-negative Graves sera contained readily detectable anti-tpoab (individual values were 85, 5, 6, 4, 115, 23, 12, 12, and 1 units/ml), emphasizing the lack of interference of anti-tpoab in the anti-tgab assay. Also, neither anti-nuclear antibodies nor rheumatoid factor crossreacted in the anti-tgab or anti-tpoab assays (data not shown). Besides being highly specific, the new assays were precise, with interassay coefficients of variation of 3.2% for anti-tpoab and 5.7% with anti-tgab. Furthermore, they showed highly significant correlations with the respective tanned red cell test (Figure 5), ELISA, and solid-phase assays based on 251-labeled Protein A. However, the new systems were an order of magnitude more sensitive, and all sera studied that were positive by tanned red cell test, ELISA, or labeled Protein A assay were clearly positive by direct assay. In common with other methods, the new assay for anti- TgAb is likely to underestimate the concentrations of anti-tgab when relatively high concentrations of endogenous Tg are present in the test serum sample. The high sensitivity and precision of the new assay method enabled definitive detection and measurement of both anti-tpoab and anti-tgab in sera that were negative by other techniques. In particular, anti-tpoab were detectable in the some Graves sera by the new assay but not by other techniques (Figure 4 and 5). Furthermore, four of 13 Hashimoto sera were negative for anti-tgab by tanned red cell test but all 13 were positive by direct assay. Overall, therefore, our studies indicate that the direct interaction between labeled Tg and TPO can be used to characterize and quantifr thyroid autoantibodies in serum. The assays have many advantages over existing procedures, including high sensitivity and specificity, good precision, and easy handling. The very high sensitivity of the system may well be of particular value in detecting low concentrations of antibody in individuals susceptible to the development of autoimmune thyroid disease; further studies in this area are in progress. References 1. Bird T, StephensonJ. Evaluation of a tanned red cell technique for thyroid microsomal antibodies. J Clin Pathol 1973;26: Schardt CW, McLachlan SM, Matheson J, Rees Smith B. An enzyme-linked immunoassay for thyroid microsomal antibodies. J Immunol Methods 1982;55: Furmaniak J, Bradbury J, Rees Smith B. Antibodies to membrane antigens in autoimmune thyroid disease. Acta Endocrinol 1987;116: RufJ, Czarnocka B, Ferrand M, Doullais F, Carayon P. Novel routine assay of thyroperoxidase autoantibodies. Clin Chem 1988;34: Southgate K, Creagh F, TeeceM, Kingswood C, ReesSmith B. A receptor assay for the measurement of TSH receptor antibodies in unextracted serum. Clin Endocrinol 1984;2: Kajita Y, Morgan D, Parkas AB, Rees Smith B. Labelling and immunoprecipitation of thyroid microsomal antigen.febs Lett 1985;187: Czarnocka B, Ruf J, Ferrand M, Carayon P, Lissitzky S. Purification of the human thyroid peroxidase and its identification as the inicrosomal antigen involved in aut.oimmune thyroid diseases. FEBS Lett 1985;19: Ohtaki S. Kotani T, Nakamura Y. Characterization of human thyroid peroxidase purified by monoclonal antibody-assisted chromatography. J Clin Endocrinol Metab 1986;63: Hosoya T, Morrison M. The isolation and purification of thyroid peroxidase. J Biol Chem 1967;242: Bradford MM. A rapid and sensitive method for the quantitation of microgram quantities of protein utilizing the principle of protein-dye binding. Anal Biochem 1976;72: Van Herle AJ, Uller RP, Matthews NL, Brown J.Radioimmunoassayfor measurementofthyroglobulin in human serum. J Clin Invest 1973;52: Derrien Y, Michel R, Roche J.Recherchessur Iapreparationet lea proprietes de Ia thyroglobuline pure. I. Biochim BiophysActs 1948;2: Fraker PJ, Speck JC. Protein and cell membrane iodinations with a sparingly soluble chioramide 1,3,4,6-tetrachloro-3a, 6a-diphenyl-glucoluril. Biochem BiophysRes Commun 1978;8: Greenwood PC, Hunter WM, Glover JS. The preparationof 311-labelledhuman growth hormone of high specific radioactivity. Biochem J 1963;89: Scatchard G. The attraction of proteins for small molecules and ions. Ann NY Aced Sci 1949;51: Fukuma N, McLachlan SM, Petersen VB, et a!. Human thyroglobulin autoantibodies of subclasses IgG2 and IgG4 bind to different epitopes on thyroglobulin. Immunology 1989;67: Petersen VB, Fukuma N, McLachlan SM, et al. A humanmouse hybridoma which secretes monoclonal thyroglobulinautoantibody with properties similar to those of the donor patient s serum autoantibody. Autoimmunity 1989;14: CLINICAL CHEMISTRY, Vol. 35, No. 9, 1989