Printer Friendly

Free thyroxine measured by equilibrium dialysis and nine immunoassays in sera with various serum thyroxine-binding capacities.

Despite the predominant role of thyrotropin measurements in the assessment of thyroid status, free thyroxine ([FT.sub.4]) measurements remain useful either when thyrotropin determination is not conclusive or when a diagnosis of thyroid disease must be confirmed (1). Because it represents only a minute fraction (0.02%) of total [T.sub.4] ([TT.sub.4]), [FT.sub.4] is more difficult to measure (2). Direct equilibrium dialysis (ED) methods are considered analytically accurate (3) and are the methods against which others are compared (4). Compared with ED, other [FT.sub.4] immunoassays may show significant biases related to protein-bound [T.sub.4] or to the serum [T.sub.4]-binding capacity (sBC: concentration x affinity of binding proteins) (4-6). We assume assays are calibrated to have roughly the same euthyroid range in samples with normal sBC, and we expect that markedly negative biases may be observed in samples with low sBC and that smaller positive biases may be observed in samples with high sBC (7). The aim of our study was to determine, in clinical samples from euthyroid patients classified into three groups as a function of their low, normal, or high sBC, the bias between [FT.sub.4] measured with ED and that measured with nine frequently used immunoassays. We also studied the specificity of each assay method and the concordance of immunoassays with ED.

[FT.sub.4] was determined with the Nichols ED/RIA assay (Nichols Institute Diagnostics) and the following nine immunoassays: Elecsys (EL) from Roche Diagnostics, VIDAS (VD) from bioMerieux, Vitros ECi (VT) from Ortho-Clinical Diagnostics, GammaCoat 2-step RIA (GC) from DiaSorin, Immulite (IM) from Diagnostic Products Corporation (DPC), Nichols Advantage (AD), AxSYM (AX) from Abbott Diagnostic, ACS (AC) from Bayer Diagnostics, and AIA (AI) from Tosoh Bioscience. All assays were performed in compliance with the manufacturers' instructions. The sBC was calculated by dividing the [TT.sub.4] concentration determined with the EL assay by the [FT.sub.4] concentration determined by ED (8). The sBC was further assessed by measuring the concentration of thyroxine-binding globulin (TBG), the main [T.sub.4]-carrier protein, with the Ria-ghost RIA from CIS bio international, and the EL T-Uptake (TU), which estimates the number of unoccupied serum protein binding sites. TU results are normalized and should be directly related to the sBC.

To study a wide range of sBCs, we selected sera from pregnant women in the last 3 months of pregnancy (n = 29) and hospitalized patients (n = 42). All patients were euthyroid, and none had been known to have patent thyroid dysfunction in the past. Except for one patient treated with heparin, none received a treatment known to interfere with the thyroid function or [T.sub.4] measurements. The sera were classified into three groups, high-, normal-, or low-sBC, depending on whether sBC results ([TT.sub.4]/ED [FT.sub.4]), were above, within, or below the reference interval determined in ambulatory patients (4.2-8.5 nmol/pmol) (8). All of the procedures that we followed were in accordance with the Helsinki Declaration of 1975 and the subsequent 1996 amendments. The patient sera were collected for routine analysis in tubes without anticoagulant, were kept frozen at -20[degrees]C, and were analyzed shortly after thawing.

The 29 sera from pregnant women were classified into the high-sBC group [mean (SD) sBC, 13.1 (2.5) nmol/ pmol], 29 sera from hospitalized patients into the normals-BC group [5.6 (1.0) nmol/pmol], and 13 from hospitalized patients into the low-sBC group [2.7 (0.9) nmol/ pmol]. Compared with the results in the normal-sBC group [TBG, 22.2 (5.0) mg/L; TU, 0.97 (0.09)], TBG and TU were significantly decreased (P <0.01, Mann-Whitney U-test) in the low-sBC group [16.4 (4.3) mg/L and 0.80 (0.13), respectively] and were increased in the high-sBC group [59.2 (10.2) mg/L and 1.39 (0.08), respectively].

The biases determined between each immunoassay and ED as a function of sBC are reported in Fig. 1. In the normal-sBC group, we found no significant bias with the EL (mean, -1.1 pmol/L), AI (-0.9 pmol/L), VT (0.8 pmol/L), and GC (1.0 pmol/L) immunoassays (P >0.05, Wilcoxon test) and a modest but significant (P <0.01) negative bias with the AC (-2.5 pmol/L), IM (-2.9 pmol/L), AX (-3.2 pmol/L), VD (-3.3 pmol/L), and AD (-3.7 pmol/L) immunoassays. The bias was more marked when the sBC was in the lower part of the reference interval (4.2-5.0 nmol/pmol). In the high-sBC group, except with the VT and AC assays, which showed no bias, we observed significant but very modest biases, negative for AX (-0.8 pmol/L) and AD (-1.1 pmol/L), and positive for the EL (1.0 pmol/L), VD (1.1 pmol/L), GC (1.4 pmol/L), AI (1.5 pmol/L), and IM (1.7 pmol/L) assays. A moderate positive bias can be related to the high sBCs of these samples, but a negative bias was unexpected. In the low-sBC group, all methods showed a significant and marked negative bias increasing in the following order: VT (-8.0 pmol/L), GC (-14.9 pmol/L), AI (-15.8 pmol/L), AC (-17.2 pmol/L), EL (-17.4 pmol/ L), AD (-17.5 pmol/L), AX (-20.4 pmol/L), VD (-20.7 pmol/L), and IM (-21.3 pmol/L).

All patients of this study were considered as euthyroid, and their [FT.sub.4] results were expected to be within the reference interval. For each assay, the [FT.sub.4] range, number of decreased or increased results, and concordance with ED are reported in Table 1. In the normal-sBC group, concordance with ED ranged between 59% (IM) and 96% (VT). GC, AI, and EL assays, for which no significant bias was evidenced, showed a poorer concordance than did AD and AC assays, which were significantly biased. These findings can be explained in terms of different calibrations. Except for the VT assay, all of the immunoassays yielded decreased [FT.sub.4] values for some of these sera despite sBCs within the reference interval. In the high-sBC group, with all immunoassays as well as with ED, irrespective of the bias, we found a high, method-dependent number of decreased results. In the low-sBC group, assays other than ED, VT, and AD yielded some decreased results.

In the normal-sBC group, the biases, when significant, were modest. This was expected because [FT.sub.4] assays are calibrated against the ED method, using sera samples from ambulatory patients with normal sBC. Except for the VT assay, however, we observed some decreased values, which lowered the concordance with ED. Therefore, the sera of not severely ill hospitalized patients with normal sBC should be included in the panel used to establish the reference interval. Alternatively, specific reference intervals for ambulatory and hospitalized individuals could be better suited. This applies in particular to methods yielding a not inconsiderable number of decreased values (i.e., IM, EL, VD, AI, and AX).

In the high-sBC group, the biases, when significant, were also modest. Contrary to previous reports (10, 11), and yet in agreement with other observations (6, 8, 12-15), subnormal values, whether measured with ED or any other immunoassay [FT.sub.4] method, are not uncommon in the last months of pregnancy. Moreover, the maximum value was clearly and systematically below the upper limit found in nonpregnant women. These findings underline the absolute necessity to consider assay-specific [FT.sub.4] reference intervals for women in the last months of pregnancy, to be in a position to diagnose not only hypothyroidism but also hyperthyroidism in a reliable way and to adjust an appropriate [T.sub.4] treatment.

In the low-sBC group, we found a significant bias with all immunoassays, VT included, contrary to what has been reported previously (8). This finding may be related to some very high ED results (up to 65.9 pmol/L). Extremely low doses of heparin release lipase activity into the plasma and can thereby cause artifactual increases of serum [T.sub.4] concentrations as measured by ED (16-19). None of the patients in this group was known to be treated with heparin. In hospitalized patients, however, heparin is frequently used for multiple blood samplings. Except for the VT and AD [T.sub.4] immunoassays, decreased values were also observed. Albumin, added to the assay ingredients to buffer the effects of increased amounts of nonesterified fatty acids that develop in serum in vitro, may have induced a negative bias (4,20). Contrary to VT reagents (21), EL (19) and AX (4) reagents probably contain albumin. Our results confirm the great variability in [FT.sub.4] measurements in hospitalized patient sera with low sBC values (11,19,22). As with ED, increased results may be observed, but contrary to ED, decreased values were yielded by most immunoassays. In these sera, some methods (ED, VT, GC) yielded preferentially increased results; others, such as IM and to a lesser extent VD, AX, EL, and AI, preferentially decreased results.


In conclusion, although methodologies have somewhat improved, as far as commonly used immunoassays are concerned, it remains absolutely necessary to consider the assay method to correctly interpret [FT.sub.4] results in pregnant women when these are decreased or lie in the upper zone of the reference interval for nonpregnant women. The same goes for hospitalized patients, regardless of whether increased or decreased results have been obtained. This implication is particularly important for sera with low sBC (severely ill patients) but also seems valid, at least as far as some immunoassays are concerned, for sera from hospitalized patients whose sBC lies in the lower part of reference interval. Clinicians should be aware of the method used to determine [FT.sub.4] because the evaluation of the effect of pregnancy, as well as of severe or even mild non-thyroidal illness, on [FT.sub.4] results varies as a function of the various immunoassays.

We are indebted to A. Beaudonnet (H6tel Dieu, Lyon, France) for carrying out ACS:180 determinations and for providing serum samples from hospitalized patients; to S. Doffoel and N. Labouret (MGEN, Strasbourg, France) for performing VT ECi determinations; to G. Forzy (Hopital Saint Philibert, Lomme, France), in charge of AI determinations, and to C. Massart (Hopital Pontchaillou, Rennes, France) for AX determinations. We thank Roche Diagnostics (Meylan, France), bioMerieux (Marcy l'Etoile, France), Ortho-Clinical Diagnostics (Issy Les Moulineaux, France), DiaSorin (Antony, France), Diagnostic Products Corporation (La Garenne-Colombes, France), Nichols Institute Diagnostics (Paris, France), Abbott Diagnostic (Rungis, France), Bayer Diagnostics (Puteaux, France), and Tosoh Bioscience (Saran, France) for supplying reagents and some of the dedicated platforms free of charge. We are also grateful to F. Gasser for invaluable help and to N. Heider for reviewing the English in the manuscript; and we acknowledge the financial support of the Hopitaux Universitaires de Strasbourg.


(1.) Spencer C, Eigen A, Shen D, Duda M, Qualls S, Weiss S, et al. Specificity of sensitive assays of thyrotropin (TSH) used to screen for thyroid disease in hospitalized patients. Clin Chem 1987;33:1391-6.

(2.) Ekins R. Analytic measurements of free thyroxine [Review]. Clin Lab Med 1993;13:599-630.

(3.) Nelson JC, Weiss RM, Wilcox RB. Underestimates of serum free thyroxine ([T.sub.4]) concentrations by free [T.sub.4] immunoassays. J Clin Endocrinol Metab 1994;79:76-9.

(4.) Christofides ND, Wilkinson E, Stoddart M, Ray DC, Beckett GJ. Serum thyroxine binding capacity-dependent bias in an automated free thyroxine assay. J Immunoassay 1999;20:201-21.

(5.) Wang R, Nelson JC, Weiss RM, Wilcox RB. Accuracy of free thyroxine measurements across natural ranges ofthyroxine binding to serum proteins. Thyroid 2000;10:31-9.

(6.) Sapin R. Serum thyroxine binding capacity-dependent bias in five free thyroxine immunoassays: assessment with serum dilution experiments and impact on diagnostic performance. Clin Biochem 2001;34:367-71.

(7.) Christofides ND. Free analyte immunoassay. In Wild D, ed. The immunoassay handbook, 2nd ed. London: Nature Publishing Group, 2001:61-77.

(8.) Christofides ND, Wilkinson E, Stoddart M, Ray DC, Beckett GJ. Assessment of serum thyroxine binding capacity-dependent biases in free thyroxine assays. Clin Chem 1999;45:520-5.

(9.) D'Herbomez M, Forzy G, Gasser F, Massart C, Beaudonnet A, Sapin R. Clinical evaluation of nine free thyroxine assays: persistent problems in specific physiopathological situation. Clin Chem Lab Med 2003;41:in press.

(10.) Frantz CR, Dagogo-Jack S, Ladenson JH, Gronowski AM. Thyroid function during pregnancy. Clin Chem 1999;45:2250-8.

(11.) Stockigt JR. Free thyroid hormone measurement: a critical appraisal [Review]. Endocrinol Metab Clin North Am 2001;30:265-89.

(12.) Roti E, Gardini E, Minelli R, Bianconi L, Flisi M. Thyroid function evaluation by different commercially available free thyroid hormone measurement kits in term pregnant women and their newborns. J Endocrinol Invest 1991;14: 1-9.

(13.) Van Blerk M, Smitz J, Rozenski E, Mees M, Roelandt P, Laermans L, et al. Four radioisotopic immunoassays of free thyroxine compared. Ann Clin Biochem 1996;33:335-43.

(14.) McElduff A. Measurement of free thyroxine ([T.sub.4]) in pregnancy. Aust NZ J Obstet Gynaecol 1999;39:158-61.

(15.) Panesar NS, Li CY, Rogers MS. Reference intervals for thyroid hormones in pregnant Chinese women. Ann Clin Biochem 2001;38:329-32.

(16.) Liewendahl K. Thyroid function tests: performance and limitations of current methodologies. Scand J Clin Lab Invest 1992;52:435-45.

(17.) Jaume JC, Mendel CM, Frost PH, Greenspan FS, Laughton CW. Extremely low doses of heparin release lipase activity into the plasma and can thereby cause artifactual elevations in the serum free thyroxine concentration as measured by equilibrium dialysis. Thyroid 1996;6:79-83.

(18.) Stevenson HP, Archbold GPR, Johnston P, Young IS, Sheridan B. Misleading serum free thyroxine results during low molecular weight treatment. Clin Chem 1998;44:1002-7.

(19.) Sapin R, Schlienger JL, Gasser F, Noel E, Lioure B, Grunenberger F, et al. Intermethod discordant free thyroxine measurements in bone marrow-transplanted patients [Technical Brief]. Clin Chem 2000;46:418-22.

(20.) Midgley JEM. Direct and indirect free thyroxine assay methods: theory and practice [Review]. Clin Chem 2001;47:1353-63.

(21.) Saw S, Sethi S, Aw TC. Technical evaluation of thyroid assays on the Vitros ECi [Technical Brief]. Clin Chem 1999;45:578-80.

(22.) Langton JE, Brent GA. Non thyroidal illness syndrome: evaluation of thyroid function in sick patients [Review]. Endocrinol Metab Clin North Am 2002;31:159-72.

Remy Sapin [1] and Michele d'Herbomez [2] [[1] Laboratoire Universitaire de Biophysique, Unite d'Analyses Endocriniennes, Universite Louis Pasteur (ULP)/Centre National de la Recherche Scientifique (CNRS) Unite Mixte de Recherche (UMR) 7004, Faculte de Medecine, 67085 Strasbourg Cedex, France; [2] Service Central de Medecine Nucleaire, Hopital Salengro, Centre Hospitalo-Universitaire Regional (CHRU), 59037 Lille Cedex, France; * address correspondence to this author at: Institut de Physique Biologique, Faculte de Medecine, F-67085 Strasbourg Cedex, France; fax 33-3-90-24-40-57, e-mail]
Table 1. FT4 measured with ED and nine immunoassays in sera with normal
sBC (4.2 nmol/pmol <sBC <8.5 nmol/pmol), high sBC (>8.5 nmol/pmol),
or low sBC (<4.2 nmol/pmol).


Reference interval, pmol/L 10.3-34.7 (a) 12.8-23.4 (b)
Normal sBC (SD), 5.6 (1.0) nmol/pmol
 (n = 29)
 Interval, (c) pmol/L 8.6-33.6 10.0-24.6
 n decreased (d)/n increased (e) 1/1 8/1
 Concordance with ED, (f) % 72

High sBC (SD), 13.1 (2.5) nmol/pmol
 (n = 29)
 Interval, (c) pmol/L 7.5-16.7 9.4-14.5
 n decreased (d)/n increased (e) 10/1 18/0
 Concordance with ED, (f) % 66

Low sBC (SD), 2.7 (0.9) nmol/pmol
 (n = 13)
 Interval, (c) pmol/L 13.6-65.9 7.8-24.5
 n decreasedd/n increased (e) 0/5 3/1
 Concordance with ED, (f) % 31


Reference interval, pmol/L 10.3-21.3 (b) 10.2-28.5 (b)
Normal sBC (SD), 5.6 (1.0) nmol/pmol
 (n = 29)
 Interval, (c) pmol/L 5.1-21.7 10.2-26.0
 n decreased (d)/n increased (e) 7/1 0/0
 Concordance with ED, (f) % 76 96

High sBC (SD), 13.1 (2.5) nmol/pmol
 (n = 29)
 Interval, (c) pmol/L 8.9-17.2 7.4-14.1
 n decreased (d)/n increased (e) 4/1 12/1
 Concordance with ED, (f) % 83 79

Low sBC (SD), 2.7 (0.9) nmol/pmol
 (n = 13)
 Interval, (c) pmol/L 5.5-22.0 13.5-40.8
 n decreasedd/n increased (e) 3/1 0/5
 Concordance with ED, (f) % 31 69


Reference interval, pmol/L 12.2-24.5 (b) 11.6-23.4 (b)
Normal sBC (SD), 5.6 (1.0) nmol/pmol
 (n = 29)
 Interval, (c) pmol/L 9.0-26.4 6.3-24.3
 n decreased (d)/n increased (e) 5/1 12/1
 Concordance with ED, (f) % 83 59

High sBC (SD), 13.1 (2.5) nmol/pmol
 (n = 29)
 Interval, (c) pmol/L 9.5-16.0 10-15.6
 n decreased (d)/n increased (e) 13/0 6/1
 Concordance with ED, (f) % 76 83

Low sBC (SD), 2.7 (0.9) nmol/pmol
 (n = 13)
 Interval, (c) pmol/L 8.8-27.1 4.4-20.3
 n decreasedd/n increased (e) 1/3 5/1
 Concordance with ED, (f) % 62 31


Reference interval, pmol/L 8.1-21.5 (b) 10.3-20.1 (b)
Normal sBC (SD), 5.6 (1.0) nmol/pmol
 (n = 29)
 Interval, (c) pmol/L 6.4-20.4 7.1-22.0
 n decreased (d)/n increased (e) 3/1 6/1
 Concordance with ED, (f) % 93 79

High sBC (SD), 13.1 (2.5) nmol/pmol
 (n = 29)
 Interval, (c) pmol/L 6.9-14.2 7.8-12.2
 n decreased (d)/n increased (e) 7/1 14/0
 Concordance with ED, (f) % 76 72

Low sBC (SD), 2.7 (0.9) nmol/pmol
 (n = 13)
 Interval, (c) pmol/L 12.1-27.2 5.9-20.9
 n decreasedd/n increased (e) 0/1 3/1
 Concordance with ED, (f) % 69 31


Reference interval, pmol/L 10.3-20.5 (b) 10.7-25.8 (b)
Normal sBC (SD), 5.6 (1.0) nmol/pmol
 (n = 29)
 Interval, (c) pmol/L 7.1-23.6 7.4-27.4
 n decreased (d)/n increased (e) 4/1 7/1
 Concordance with ED, (f) % 86 76

High sBC (SD), 13.1 (2.5) nmol/pmol
 (n = 29)
 Interval, (c) pmol/L 8.7-13.4 8.2-16
 n decreased (d)/n increased (e) 5/1 7/1
 Concordance with ED, (f) % 83 90

Low sBC (SD), 2.7 (0.9) nmol/pmol
 (n = 13)
 Interval, (c) pmol/L 7.1-21.2 7.2-26.6
 n decreasedd/n increased (e) 1/2 3/1
 Concordance with ED, (f) % 58 50

(a) From the assay package insert and Ref. (4), range observed in
263 nonpregnant healthy adults with one outlier deleted at each

(b) Determined in a multicenter study from the results of 152
control sera of ambulatory euthyroid patients (2.5th-97.5th
percentile) (9).

(c) Interval, minimum and maximum [FT.sub.4] values with each

(d) n decreased, number of results below the lower limit of the
reference interval of each method.

(e) n increased, number of results above the upper limit of the
reference interval of each method.

(f) Concordance between ED and each immunoassay expressed as the
percentage of concordant results.
COPYRIGHT 2003 American Association for Clinical Chemistry, Inc.
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2003 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Title Annotation:Technical Briefs
Author:Sapin, Remy; d'Herbomez, Michele
Publication:Clinical Chemistry
Date:Sep 1, 2003
Previous Article:Characterization of the bell polymorphism in the glucocorticoid receptor gene.
Next Article:High Mobility Group protein 1(HMGB1) quantified by ELISA with a monoclonal antibody that does not cross-react with HMGB2.

Terms of use | Privacy policy | Copyright © 2020 Farlex, Inc. | Feedback | For webmasters