Clinical Chemistry 48:1
102–107 (2002)
Importance of the Detection Method for
Thyroglobulin Antibodies for the Validity of
Thyroglobulin Measurements in Sera from
Patients with Graves Disease
Catherine Massart1* and Didier Maugendre2
Background: The use of recovery tests has been pro-
posed to disclose interferences from anti-thyroglobulin
antibodies (TgAbs) in thyroglobulin (Tg) assays. We
studied the value of a recovery test in Tg measurement
by a new commercial IRMA.
Methods: Blood samples were collected from 153 pa-
tients with untreated Graves disease. Tg and TgAbs
were measured by IRMA and RIA, respectively
(Dynotest Tg-plus and Dynotest anti-Tgn; Brahms Di-
agnostica). The recoveries of added amounts of Tg were
calculated for each serum.
Results: TgAbs were detected in 72 of the 153 patients
(47%). The recovery test results for the 81 TgAb-negative
sera (median, 101%; range, 80 –115%) were identical to
the results for the 91 controls (median, 102%; range,
80 –124%). By contrast, significantly lower recovery test
results were observed for the 72 TgAb-positive sera
(median, 79%; range, 60 –103%; Z ⴝ ⴚ8.363; P <0.0001).
In the 34 of the 72 TgAb-positive sera with a normal
recovery test, Tg concentrations were significantly lower
(median Tg, 13.6 ␮g/L; range, 1.1–360 ␮g/L) than those
measured in the TgAb-negative sera (median, 107 ␮g/L;
range, 1.2–700 ␮g/L; Z ⴝ ⴚ3.797; P <0.0001).
Conclusions: Tg values were decreased in TgAb-posi-
tive sera even when the results of the recovery tests were
normal. This test should not be used alone to determine
the validity of a serum Tg measurement in Graves
disease.
© 2002 American Association for Clinical Chemistry
1
Laboratoire de Génétique Moléculaire et Hormonologie et 2 Unité
d’Endocrinologie, CHU de Pontchaillou, Rue H. Le Guilloux, 35043 Rennes
Cedex, France.
*Author for correspondence. Fax 02-99-28-41-45; e-mail catherine.
massart@chu-rennes.fr.
Received June 22, 2001; accepted October 8, 2001.
102
Endocrinology and
Metabolism
Graves disease is an autoimmune disorder characterized
by hyperthyroidism attributable to thyroid-stimulating
hormone (TSH)3 receptor autoantibodies with thyroid-
stimulating activity (1–3 ). The measurement of TSH re-
ceptor-binding antibodies in serum can be used to follow
the course of Graves disease (4, 5 ). However, this single
marker cannot predict the outcome for individual Graves
patients (6 ) even when TSH receptor-binding antibodies
are assayed with a sensitive radioreceptor assay (7 ).
Because the serum thyroglobulin (Tg) concentration re-
flects the magnitude of TSH receptor stimulation (8 ), it
has been reported that measurement of serum Tg can
provide useful information for monitoring the course of
Graves disease and for managing therapeutic treatment
(9 –12 ).
Measurement of Tg in serum is hampered by the
presence of circulating interfering factors, especially Tg
antibodies (TgAbs), which are frequent in Graves disease
and which affect the reliability of Tg assays (8, 13 ).
Because most current RIA methods do not reliably mea-
sure serum total Tg, IRMAs using monoclonal TgAbs that
do not cross-react with endogenous TgAbs have been
developed (14, 15 ). These IRMAs do not necessarily pro-
vide any substantial advantage over a conventional poly-
clonal IRMA in detecting Tg in TgAb-positive sera (16 ).
It has been suggested (17 ) that quantitative assessment
of Tg in TgAb-positive sera may be achieved by determi-
nation of the recovery of known amounts of Tg added to
TgAb-positive samples. We used a new commercially
available IRMA incorporating both polyclonal and mono-
clonal TgAbs to test TgAb interference and the value of
the recovery test in sera from nontreated Graves patients.
3
Nonstandard abbreviations: TSH, thyroid-stimulating hormone; Tg, thy-
roglobulin; and TgAb, thyroglobulin antibody.
 Clinical Chemistry 48, No. 1, 2002 103

Materials and Methods where A is the supplemented serum sample, and B is the
patients unsupplemented serum sample.
The study involved 153 patients (125 women and 28 men; All determinations were performed in duplicate. The
median age, 42 years; range, 17– 65 years) with Graves Tg standard was calibrated with the International Stan-
disease diagnosed from typical clinical signs: hyperthy- dard Certified Reference Material 457 in the following
roidism, vascular and homogeneous goiter, occasional ratio: 1 ng of Certified Reference Material was equal to 0.5
exophthalmos, increased free thyroid hormone concentra- ng of the Dynotest Tg-plus standard.
tions (free triiodothyronine ⬎8.9 pmol/L and free thyrox-
ine ⬎23.4 pmol/L), and undetectable TSH (⬍0.05 mIU/ statistical analysis
L). Patients showing toxic nodules were excluded. Quantitative variables were analyzed using nonparamet-
Ninety-one apparently healthy euthyroid individuals ric tests. The Mann–Whitney test was used to compare the
(82 women and 9 men; median age, 41 years; range, 20 – 60 variables in the different groups. Correlation analyses
years) served as controls. These individuals did not were performed using the Spearman test.
smoke and had no goiter or no personal or family history
of thyroid disease, and thyroid autoantibodies were neg- Results
Tg method
ative.
The imprecision of the assay (CV) was ⬍3% at 3–335 ␮g/L
We assayed Tg and TgAbs in the 153 Graves patients
(Table 1). The functional assay sensitivity, defined as an
before treatment. Blood samples were collected in antico-
interassay CV ⬍20% during a 6-month period (8 ), was 0.2
agulant-free tubes and centrifuged at 1000g for 10 min at
␮g/L. Dilution curves for serum samples paralleled the
4 °C. Sera were decanted for storage at ⫺20 °C until assay.
calibration curve (Fig. 1).
Among 91 samples from healthy volunteers, 4 had
Tg assay TgAb concentrations ⱖ60 kilounits/L. These four samples
Tg was measured by IRMA using a new commercial were eliminated from the estimation of reference values.
reagent set (Dynotest Tg-plus; Brahms Diagnostica, Ber- The Tg values for the remaining 87 serum samples
lin, Germany). The method as described by the manufac- showed a logarithmic distribution (mean, 7.6 ␮g/L;
turer is as follows: Standard or patient’s serum (100 ␮L) is range, 2.45–22 ␮g/L).
pipetted into test tubes coated with polyclonal TgAb. The
tubes are incubated for 18 h at room temperature and then evaluation of recovery
washed twice with 2 mL of washing solution. The tubes For 30 sera with low Tg values (⬍3 ␮g/L), the recovery
are turned upside down on blotting paper for at least 10 test was performed by adding 1 ␮g/L Tg; for other sera,
min. After the tubes are again turned right side up, 200 ␮L we added 50 ␮g/L. For the 87 control samples considered
of 125I-labeled monoclonal TgAb is added. The tubes are negative for TgAbs (⬍60 kilounits/L), the median recov-
incubated for 2–3 h at room temperature with shaking ery was 102% (range, 80 –124%) compared with 82%
(300 – 400 rpm), after which they are washed twice with 2 (range, 65–98%) for the 4 TgAb-positive samples.
mL of washing solution and then turned upside down for In the 153 sera from patients with Graves disease,
at least 10 min on blotting paper. The radioactivity of each median recovery was 90% (range, 62–113%), which was
tube is then measured. significantly lower than that obtained for the controls
Sera with high Tg concentrations were diluted with the (Z ⫽ ⫺5.606; P ⫽ 0.0001). Recovery was ⬍80% for 39
buffer included in the reagent set. Two dilutions (undi- (25%) of the 153 sera.
luted and 1:10) were prepared for each specimen with an For 81 samples considered negative for TgAbs (⬍60
increased Tg concentration 10-fold above the upper limit kilounits/L), median recovery was 101% (range, 80 –
of the assay range (⬎220 ␮g/L). Interference from auto- 115%), which was identical to the recovery for control sera
antibodies was assessed by direct measurement of TgAb (Z ⫽ ⫺1.134; P ⬎0.05). By contrast, significantly lower
with a RIA (Dynotest anti-Tgn; Brahms Diagnostica); all
values ⬍60 kilounits/L were considered negative. The Tg
IRMA method also contained an estimate, by routine Table 1. Precision of the Tg assay.a
determination, of recovery that was carried out by adding Intraassay Interassay
the 50 ␮g/L Tg calibrator to each serum tested. The Mean ⴞ SD, ␮g/L CV, % n Mean ⴞ SD, ␮g/L CV, % n
recovery test was also performed by adding 1 ␮g/L Tg to 1.75 ⫾ 0.13 7.3 12 3.36 ⫾ 0.096 2.9 16
several sera containing a low concentration of Tg (⬍3 12.3 ⫾ 0.2 1.7 12 12.5 ⫾ 0.29 2.3 16
␮g/L). Recovery (%) was calculated as: 43.0 ⫾ 0.66 1.5 12 43.5 ⫾ 1 2.3 16
335.0 ⫾ 7.6 2.3 12 332.0 ⫾ 6.9 2.1 16
␮g Tg/L (A) a
Intra- and interassay variations were calculated from duplicate Tg measure-
⫺ ␮g Tg/L (B) ments of four pooled sera (n represents the number of assays in the same series
⫻ 100 for intraassay precision and the number of series for interassay precision).
50 (or 1) ␮g Tg/L
104 Massart and Maugendre: Thyroglobulin Assay in Graves Disease

Fig. 1. Dilution test for Tg assay with three sera (⫹, ‚, and E).

recoveries were observed in the 72 samples with TgAb

concentrations ⱖ60 kilounits/L (median, 79%; range, 60 –
103%; Z ⫽ ⫺8.363; P ⬍0.0001).
Recovery values and TgAb activities were significantly
correlated (r ⫽ ⫺0.602; P ⬍10⫺4). Except for one sample,
all sera with a negative TgAb activity had a normal
recovery (ⱖ80%; Fig. 2A). In 38 (53%) of the 72 TgAb-
positive sera, a recovery ⬍80% was observed, whereas 34
(47%) of the 72 sera had a normal recovery test (Fig. 2B).
Tg values in the 72 TgAb-positive sera (median, 7
␮g/L; range, 0.16 –277 ␮g/L) were significantly lower
than those obtained in the 81 TgAb-negative sera (medi-
an, 114 ␮g/L; range, 1.2–700 ␮g/L; Z ⫽ ⫺4.788; P
⬍0.0001; Fig. 3A). Moreover, in the 34 of the 72 TgAb-
positive sera with a normal recovery test (ⱖ80%), Tg
values were also significantly lower (median Tg, 13.6
␮g/L; range, 1.1–360 ␮g/L) than those obtained in the
TgAb-negative samples with a correct recovery (median,
107 ␮g/L; range, 1.2–700 ␮g/L; Z ⫽ ⫺3.797; P ⬍0.0001;
Fig. 3B). Fig. 2. Recovery test results for 81 TgAb-negative sera (A) and 72
Thirty-one (79%) of the 39 sera with a low recovery test TgAb-positive sera (B).
result had Tg values ⬍22 ␮g/L (Fig. 4).

Discussion attempt to overcome TgAb interference (14, 15, 24, 25 ).

Tg assays in serum are used in the follow-up of patients
with thyroid diseases such as thyroid carcinoma and
Graves disease (12, 18, 19 ). Tg assays also contribute to
the diagnosis of factitious hyperthyroidism (20 ) and to
the detection of functioning tissue in cases of congenital
hypothyroidism (21 ). Investigators using competitive RIA
methods based on polyclonal antibodies have reported
interference caused by TgAbs (22, 23 ). More recently,
IRMAs using monoclonal TgAbs that do not cross-react
with endogenous TgAbs have been developed in an
However, the use of a monoclonal antibody IRMA for
serum Tg, although less susceptible to in vitro TgAb
interference, would not necessarily provide any substan-
tial advantage (16 ). These interferences have led to an
underestimation of Tg concentrations as well as decreased
recovery of a known concentration of Tg added to the
serum (17 ). The commercial availability of a new IRMA
method that uses both a polyclonal coated TgAb and a
labeled monoclonal TgAb as a tracer enabled us to study
TgAb interference in the assay.
 Clinical Chemistry 48, No. 1, 2002
Fig. 3. Tg concentrations in TgAb-negative and -positive sera.
(A), Tg concentrations in 81 TgAb-negative sera (1) and 72 TgAb-positive sera (2).
(B), Tg concentrations in sera with a normal recovery test result [80 TgAb-
negative sera (1) and 34 TgAb-positive sera (2)].
The Tg concentrations obtained in controls agreed with
other reports on Tg reference intervals (26, 27 ). To deter-
mine the extent of this interference, we evaluated the
TgAb concentrations and the recovery of added Tg. For
TgAb measurement, we used a very sensitive RIA be-
cause it has been shown that TgAbs should be measured
quantitatively in every specimen by a specific immunoas-
say and not an insensitive qualitative hemagglutination
test (8, 26 ). We chose 80% recovery of added Tg as the
cutoff for a normal test because all except one of the
samples tested that had no measurable TgAbs gave a
recovery value ⬎80%. This cutoff is different from that
recommended by the manufacturer, who considers all
values ⱖ70% as normal. However, our lower limit of
correct recovery was identical to that obtained in the
literature (26, 27 ). The poor recovery for the one sample
may be attributable to other interfering factors in serum,
as suggested previously (26 ). We found no recovery rates
⬎130%, as reported in some studies and for which no
clear explanation has been put forward (26, 28 ). The
105
Fig. 4. Histogram showing distribution of Tg concentrations in 153
sera.
e, samples with Tg concentrations ⬍22 ␮g/L; o, samples with Tg concentra-
tions ⱖ22 ␮g/L.
inability of the assay to detect concentrations ⬎80% of
added Tg in 47% of the samples containing TgAbs indi-
cates that some interference from TgAbs was present in
the assay. This percentage was similar to the 41% ob-
tained by Erali et al. (27 ). The proportion of sera with a
low recovery test result (25%) was higher than that
obtained with monoclonal IRMAs. However, TgAbs are
found more frequently in patients with autoimmune
diseases than in patients with differentiated thyroid can-
cer, as reported in extensive studies in the literature
(26, 28 ).
Our data support the findings (17, 28, 29 ) that TgAb
interference leads to decreased Tg values because the
median Tg was lower in TgAb-positive sera than in sera
without TgAbs. Thus, an underestimation of the Tg
concentrations is possible in the presence of TgAbs. This
underestimation of Tg may produce false negatives dur-
ing cancer monitoring. Patients with thyroid carcinoma
typically have TgAbs with broader-based epitope speci-
ficities than do patients with autoimmune thyroid dis-
eases (30 ). However, the different TgAbs found in sera
from patients with thyroid cancer and Graves disease
recognize mainly region II and, occasionally, region IV on
the human Tg molecule (31 ). Thus, TgAbs could also
interfere with Tg in sera from patients with thyroid
carcinoma.
Recovery tests should not be used in attempts to
correct for TgAb effects when measuring serum Tg. With
TgAbs, poor recovery rates spread out in the low range of
Tg values (⬍22 ␮g/L). However, 8 of the 39 sera in our
106 Massart and Maugendre: Thyroglobulin Assay in Graves Disease
study with recovery rates ⬍80% did not display low Tg
concentrations. Moreover, Tg values remained signifi-
cantly reduced in TgAb-positive sera despite normal
recovery test results. Thus, inappropriately low Tg results
were the consequence of in vitro TgAb interference, which
recovery tests sometimes failed to detect. The TgAb
measurement is a more suitable tool to disclose these
interferences. Tg concentrations should be interpreted
with respect to the presence or absence of TgAbs. These
results are in agreement with those of Spencer et al. (8 ),
who reported poor recovery test values in six Tg assays
performed with other commercial RIA or IRMA methods.
Finding a lower than expected serum Tg value in TgAb-
positive sera may indicate a truly low Tg concentration,
possibly because of accelerated Tg metabolic clearance
(16 ). Measurement of TgAbs remains an absolute prereq-
uisite for the correct interpretation of serum Tg assays.
In conclusion, with this new IRMA, Tg results were
affected by TgAbs in Graves disease sera although the
recovery test results were normal. Consequently, the Tg
recovery test is not a reliable way to detect interference in
all TgAb-positive sera; it should not be used alone to
determine the validity of a serum Tg measurement in the
monitoring of patients with Graves disease. A similar
study on sera from patients with differentiated thyroid
carcinoma would be useful.
We thank Brahms Diagnostica (Berlin, Germany) for the
gift of Dynotest Tg-plus and Dynotest anti-Tgn reagent
sets.
References
1. Smith BR, McLachlan SM, Furmaniak J. Autoantibodies to the
thyrotropin receptor. Endocr Rev 1988;9:106 –21.
2. McKenzie JM, Zackarija M. The clinical use of thyrotropin receptor
antibody measurement. J Clin Endocrinol Metab 1989;69:
1093– 6.
3. Burman KD, Baker J. Immune mechanism in Graves’ disease.
Endocr Rev 1985;6:183–232.
4. Madec AM, Laurent MC, Lorcy Y, Le Guerrier AM, Rostagnat-
Stephanutti A, Orgiazzi J, et al. Thyroid stimulating antibodies: an
aid to the strategy of treatment of Graves’ disease. Clin Endocrinol
1984;21:247–55.
5. Davies TF, Roti E, Braverman LE, DeGroot LJ. Thyroid controver-
sy—stimulating antibodies [Therapeutic controversy]. J Clin Endo-
crinol Metab 1998;83:3777– 85.
6. Feldt-Rasmussen U, Schleusener H, Carayon P. Meta-analysis
evaluation of the impact of thyrotropin receptor antibodies on long
term remission after medical therapy of Graves’ disease. J Clin
Endocrinol Metab 1994;78:98 –102.
7. Maugendre D, Massart C. Clinical value of a new TSH binding
inhibitory activity assay using human TSH receptors in the fol-
low-up of antithyroid drug treated Graves’ disease. Comparison
with thyroid stimulating antibody assay. Clin Endocrinol 2001;54:
89 –96.
8. Spencer CA, Takeuchi M, Kazarosyan M. Current status and
performance goals for serum thyroglobulin assays. Clin Chem
1996;42:164 –73.
9. Werner RS, Romaldini JH, Farah CS, Werner MC, Bromberg N.
Serum thyroid-stimulating antibody, thyroglobulin levels, and thy-
roid suppressibility measurement as predictors of the outcome of
combined methimazole and triiodothyronine therapy in Graves’
disease. Thyroid 1991;1:293–9.
10. Talbot JN, Duron F, Feron R, Aubert P, Milhaud G. Thyroglobulin,
thyrotropin and thyrotropin binding inhibiting immunoglobulins
assayed at the withdrawal of antithyroid drug therapy as predic-
tors of relapse of Graves’ disease within one year. J Clin Invest
1989;12:589 –95.
11. Kawamura S, Kishino B, Tajima K, Mashita K, Tarui S. Serum
thyroglobulin changes in patients with Graves’ disease treated
with long term antithyroid drug therapy. J Clin Endocrinol Metab
1983;56:507–12.
12. Uller RP, Van Herle AJ. Effect of therapy on serum thyroglobulin
levels in patients with Graves’ disease. J Clin Endocrinol Metab
1978;46:747–55.
13. Spencer CA, Wang C. Thyroglobulin measurement: techniques,
clinical benefits and pitfalls. Endocrinol Metab Clin North Am
1995;24:841– 63.
14. Piechaczyk M, Baldet L, Pau B, Bastide JM. Novel immunoradio-
metric assay of thyroglobulin in serum with use of monoclonal
antibodies selected for lack of cross-reactivity with autoantibod-
ies. Clin Chem 1989;35:422– 4.
15. Schlumberger M, Fragu P, Gardet P, Lumbroso J, Violot D,
Parmentier C. A new immunoradiometric assay (IRMA) system for
thyroglobulin measurement in the follow-up of thyroid cancer
patients. Eur J Med 1991;18:153–7.
16. Mariotti S, Barbesino G, Caturegli P, Marino M, Manetti L, Pacini
F, et al. Assay of thyroglobulin in serum with thyroglobulin auto-
antibodies: an unobtainable goal? J Clin Endocrinol Metab 1995;
80:468 –72.
17. Bayer MF, Kriss JP. Immunoradiometric assay for serum thyroglob-
ulin: semi-quantitative measurement of thyroglobulin in antithyro-
globulin-positive sera. J Clin Endocrinol Metab 1979;49:557– 64.
18. Van Herle AJ, Uller RP. Increased serum thyroglobulin: a marker of
metastases in differentiated thyroid carcinomas. J Clin Invest
1975;56:272–7.
19. Ozata M, Suzidi S, Miyamoto T, Liu RT, Fierro-Renoy F, DeGroot LJ.
Serum thyroglobulin in the follow-up of patients with treated
differentiated thyroid cancer. J Clin Endocrinol Metab 1994;79:
98 –105.
20. Mariotti S, Martino E, Cupini C, Lari R, Giani C, Bashieri L, et al.
Low serum thyroglobulin as a clue to the diagnosis of thyrotoxi-
cosis facticia. N Engl J Med 1982;307:410 –2.
21. Pacini F, Lari R, La Ricca P, Grasso L, Di Bartolo F, Fenzi GF, et al.
Serum thyroglobulin determinations in the differential diagnosis of
congenital hypothyroidism. J Endocrinol Invest 1984;7:29 –33.
22. Schneider AB, Pervos R. RIA of human thyroglobulin: effect of
antithyroglobulin autoantibodies. J Clin Endocrinol Metab 1978;
47:126 –37.
23. Feldt-Rasmussen U, Krogh Rasmussen A. Serum thyroglobulin
(Tg) in presence of thyroglobulin autoantibodies (TgAb). Clinical
and methodological relevance of the interaction between Tg and
TgAb in vitro and in vivo. J Endocrinol Invest 1985;8:571–9.
24. Preissner CM, Klee CG, Krco CJ. Non-isotopic “sandwich” immu-
noassay of thyroglobulin in serum by the biotin-streptavidin tech-
nique: evaluation and comparison with an immunoradiometric
assay. Clin Chem 1988;35:1794 – 8.
25. Kato R, Noguchi S, Noguchi A. Human serum thyroglobulin deter-
mination with monoclonal antibody one-step assay: minimum
interference of autoantibody. Endocrinol Jpn 1987;34:171– 8.
26. Schaadt B, Feldt-Rasmussen U, Rasmusson B, Torring H, Foder B,
Jorgensen K, et al. Assessment of the influence of thyroglobulin
(Tg) autoantibodies and other interfering factors on the use of
 Clinical Chemistry 48, No. 1, 2002
serum Tg as tumor marker in differentiated thyroid carcinoma.
Thyroid 1995;5:165–70.
27. Erali M, Bigelow RB, Meikle AW. ELISA for thyroglobulin in serum:
recovery studies to evaluate autoantibody interference and reli-
ability of thyroglobulin values. Clin Chem 1996;42:766 –70.
28. Sapin R, Gasser F, Chambron J. Recovery determination in 600
sera analyzed for thyroglobulin with a recently commercialized
IRMA kit. Clin Chem 1992;38:1920 –1.
29. Bourrel F, Hoff M, Regis H, Courrière P, Caron P. Immunometric
assay of thyroglobulin in patients with differentiated thyroid carci-
107
nomas: need for thyroglobulin recovery tests. Clin Chem Lab Med
1998;36:725–30.
30. Ruf J, Carayon P, Lissitzky S. Various expressions of a unique
antihuman thyroglobulin antibody repertoire in normal state and
autoimmune disease. Eur J Immunol 1985;15:268 –72.
31. Piechaczyk M, Bouanani M, Salhi SL, Baldet L, Bastide M, Pau B,
et al. Antigenic domains on the human thyroglobulin molecule
recognized by autoantibodies in patients’ sera and by natural
autoantibodies isolated from the sera of healthy subjects. Clin
Immunol Immunopathol 1987;45:114 –21.