Technical evaluation of the first fully automated assay for the detection of TSH [620476]

Technical evaluation of the first fully automated assay for the detection of TSH
receptor autoantibodies
Derik Hermsena,⁎, Martina Broecker-Preussb, Marco Casatic, Jordi Camara Masd, Anja Ecksteine,
Dieter Gassnerf, Josef van Heldeng, Keiko Inomatah, Jochen Jarauschf, Jürgen Kratzschi, Klaus Mannb,
Naoko Miyazakij, Miguel Angel Navarro Morenod, Tsukasa Murakamih, Heinz-Jürgen Rothk,
Jaeduk Yoshimura Nohj, Werner A. Scherbauml, Matthias Schottl
aCentral Institute for Clinical Chemistry and Laboratory Medicine, University Hospital Duesseldorf, Germany
bDepartment of Endocrinology and Division of Laboratory Research, University Hospital Essen, Germany
cDepartment of Clinical Pathology, San Gerardo Hospital, Monza, Italy
dHormone and Genetic Unit, Clinical Laboratory, Hospital Universitario de Bellvitge, Barcelona, Spain
eDepartment of Ophthalmology, University Hospital Essen, Germany
fRoche Diagnostics GmbH, Penzberg, Germany
gMedical Center for Laboratory Medicine, Mönchengladbach, Germany
hNoguchi Thyroid Clinic and Hospital Foundation, Ooita, Japan
iInstitute of Laboratory Medicine, Clinical Chemistry and Molecular Diagnostics, University Hospital Leipzig, Germany
jITO Hospital, Tokyo, Japan
kLimbach Laboratory, Heidelberg, Germany
lDepartment of Endocrinology, Diabetes and Rheumatology, University Hospital Duesseldorf, Germany
abstract article info
Article history:
Received 6 November 2008
Received in revised form 18 November 2008Accepted 18 November 2008Available online 3 December 2008
Keywords:Automated TRAb assay
M22 human monoclonal antibody
Graves' diseaseBackground: Graves' disease (GD) is mediated by autoantibodies which bind to the TSH receptor (TRAb). The
aim of the present study was to evaluate the technical performance of the first fully automated immunoassay
for TRAb detection.
Methods: The Elecsys®Anti-TSHR immunoassay utilizes a porcine TSH receptor (TSHR) and the human
thyroid stimulating monoclonal TSHR autoantibody M22.
Results: Intraassay and total imprecision CV were determined between 1.4% –14.9%, and 2.4% –28.8%,
respectively. Using the 20% CV criteria the functional sensitivity was found at 0.73 IU/L. The median CV at thecut-off (1.75 IU/L) was found to be 11%. Comparison studies with five TRAb immunoassays yielded slopes and
intercepts between 1.02 –1.48, and −0.74–0.56, respectively. Correlation coef ficients were determined
between 0.895 and 0.978. ROC plot analysis of patients with GD, patients with other thyroid disorders and
healthy controls revealed an AUC of 0.99 resulting in a sensitivity of 97% and a speci ficity of 99% at a TRAb
level of 1.75 IU/L.
Conclusion: The evaluation of the TRAb immunoassay generated homogeneous performance data and
demonstrated a high degree of comparability to established TRAb assays. The automated TRAb assayrepresents a major improvement of thyroid testing in clinical practice.
© 2008 Elsevier B.V. All rights reserved.
1. Introduction
The thyrotropin receptor (TSHR) is one of the major autoantigens
in autoimmune thyroid diseases, and autoantibodies acting as TSHR
agonists lead to clinical symptoms of Graves' disease (GD) [1].I n
clinical practice the detection of TSH receptor autoantibodies (TRAb) is
a valuable aid not only in con firming the diagnosis of GD with mild
hyperthyroidism and euthyroid Graves' ophthalmotpathy (GO) with-out goiter, but also in differentiating GD from toxic nodular goiter or
factitious thyrotoxicosis [2–5]. Additionally, it is of particularly
importance for pregnant and lactating women or patients with recent
iodine load where technetium thyroid scintigraphy is contraindicated
or not helpful [6–8]. More importantly, TRAb detection is a potentially
useful tool for outcome prediction in patients with GD [9–11]and GO
[12], respectively.
Over the last few decades, great improvement has been achieved
with regard to TRAb detection methodologies [4,5] . At present, three
major techniques to detect serum TRAb have been established. Among
them, the competitive assays are the only validated routine tests for
TRAb analysis [2,4,5,13] . Traditionally, these assays are based on theClinica Chimica Acta 401 (2009) 84 –89
⁎Corresponding author. Institute for Clinical Chemistry and Laboratory Diagnostics,
University Hospital Düsseldorf, Germany. Fax: +49 211 811 8013.
E-mail address: hermsen@med.uni-duesseldorf.de (D. Hermsen).
0009-8981/$ –see front matter © 2008 Elsevier B.V. All rights reserved.
doi:10.1016/j.cca.2008.11.025
Contents lists available at ScienceDirect
Clinica Chimica Acta
journal homepage: www.elsevier.com/locate/clinchim

ability of TRAb to inhibit TSHR binding of labelled bovine TSH.
Alternatives to bovine TSH have become available for use in
competitive assays [14,15] in the last few years.
Some years ago, Sanders et al. developed a new assay with labelled
mouse thyroid-stimulating monoclonal antibodies and TSHR binding
was shown to be inhibited in a similar way to inhibition of labelled
bovine TSH binding [14]. More recently, the same group reported an
enzyme-linked immunosorbent assay (ELISA) procedure measuring
TRAb in patient serum using a labelled human thyroid stimulating
monoclonal autoantibody M22 [15]. Up to now, two comparison
studies have been published. Both studies were conducted by Kamijo
et al. who found that the new M22-biotin-based ELISA was superior to
TSH-biotin-based second generation porcine ELISA in terms of
diagnostic sensitivity and speci ficity [16,17] . Another evaluation of
an M22-based assay format at one of the evaluation sites has,
however, revealed an insuf ficient calculated between-run coef ficient
of variation at the recommended cut-off level for positivity [18].
Currently, the first fully automated electrochemiluminescence
immunoassay (ECLIA) for the determination of TRAb based on the
inhibition of binding of the human thyroid-stimulating monoclonal
antibody (M22) has been developed and became commercially
available [19]. The aim of our multicentre study was to evaluate the
technical performance of this new automated immunoassay in routine
clinical laboratories and to compare these results with other well-
established commercially available manual TRAb assays.
2. Materials and methods
2.1. Study design
The multicentre evaluation of the Elecsys®Anti-TSHR immunoassay was performed
at nine clinical evaluation sites in Europe and Japan on fully automated MODULARANALYTICS
®E170 (7 sites) and Elecsys®2010 systems (3 sites) according to a
standardized protocol (Study No. B06 P 015). In a familiarization trial intra-assayimprecisions of provided control sera were measured. The following technical part ofthe study was divided into an initial, a main and a special performance trial,
respectively. In the initial trial, the intra-assay imprecision study was completed
using six self-prepared human pool sera. An inter-laboratory survey using five provided
human pool sera was performed subsequently to check that all laboratories obtainedequivalent test results. The main trial was started with the determination of the totalimprecision. Based on the total imprecision data of all sites, the functional sensitivity ofthe Elecsys
®Anti-TSHR immunoassay was calculated. In the following technical method
comparison study each laboratory compared the Elecsys®Anti-TSHR immunoassay
with an established routine test. Finally, in the special performance trial ,the analytical
sensitivity and the dilution linearity was studied. Additionally a lot to lot comparisonwas performed using two validation lots of the Elecsys
®Anti-TSHR assay. In parallel to
these experiments the determination of the optimal decision threshold level forpositivity by receiver-operating characteristic (ROC) analysis [20] was determined at
two evaluation sites in 457 samples from apparently healthy individuals, patients withuntreated Graves' disease (GD) and patients with other thyroid diseases than GD.
2.2. Test principle
According to the information given in the package insert the Elecsys
®Anti-TSHR
immunoassay uses a solubilized porcine TSHR which becomes immobilized tostreptavidin-coated microparticles by a biotinylated mouse monoclonal captureantibody. The capture antibody binds to the C-terminal moiety of the porcine TSHR.The assay uses a competitive test principle. Serum TRAb bound to the TSH receptor aredetected by their ability to inhibit the binding of ruthenium labelled human TSHR
stimulating monoclonal antibody (M22). The assay is designed for the Elecsys and
cobase immunoassay analyzers. The Elecsys
®Anti-TSHR immunoassay is calibrated in
IU/L against the NIBSC 90/672 standard. Calibration has been performed with every setof reagent and pretreatment rackpack and has been renewed each day of use. The testrequires a sample volume of 50 µL and total assay time of 27 min. Its measuring rangeextends from 0.3 to 40 IU/L [19].
2.2.1. Imprecision study
The intra-assay imprecision study was carried out at eight laboratories using a
control of two levels, PreciControl ThyroAB (PC THYRO Level 1 range 3.86 –5.80 IU/L,
Level 2 range 15.28 –22.92 IU/L) provided by Roche Diagnostics and six self-prepared
human pool sera at different concentrations (range 0.4 –26 IU/L). The intra-assay
imprecision was determined by 21-fold measurements in a single run on MODULARANALYTICS
®E170 (7 sites) and Elecsys®2010 systems (2 sites). Mean, standard
deviation (SD) and coef ficients of variation (CV) of each site and median CVs were
calculated.The total-imprecision study was performed according to the CLSI/NCCLS guideline
EP5, A2 using aliquots of control materials (PC THYRO Level 1 range 3.86 –5.80 IU/L, Level
2 range 15.28 –22.92 IU/L) provided by Roche Diagnostics and six self-prepared pool sera
(mean range 0.4 –26 IU/L) which were analyzed in three determinations per run and two
runs per day on 10 days ( n=60 measurements). The total-imprecision was estimated in
six evaluation sites on MODULAR ANALYTICS®E170 (4 sites) and Elecsys®2010 systems
(2 sites). Mean, SD and CVs of each laboratory and median CVs were calculated. Aliquotsof frozen ( −20 °C) human pool sera as well as aliquots of the controls were thawed
immediately before analysis as recommended by the manufacturer.
The functional sensitivity (FS) was de fined as the lowest concentration that can be
measured with an interassay CV of ≤20%. The FS was assessed using the total
imprecision data of all evaluation sites. Calculated CVs of each site were plotted vs. the
means of the measured self-prepared pool sera. The FS (CV ≤20%) was determined by
regression analysis.
2.3. Inter-laboratory survey
Eight laboratories participated in an inter-laboratory survey. TRAb levels were measured
in two runs per day up to five days using five different lyophilized pool sera (samples A –E,
target range 1.3 –19.1 IU/L ) provided by Roche Diagnostics. Min/Max and Median values of
each laboratory were compared with the two-fold standard deviation (2 SD) limits of thetarget value of each sample to check that all participating laboratories obtained comparableTRAb values on MODULAR ANALYTICS
®E170 (6 sites) and Elecsys®2010 (2 sites) systems.
2.4. Technical method comparison
Method comparison studies with the Elecsys®Anti-TSHR immunoassay were carried
out in eight participating laboratories on MODULAR ANALYTICS®E170 and Elecsys®2010
systems. The Elecsys®Anti-TSHR immunoassay was compared with a TRAb radiometric
immunoassay (DYNOtest TRAK human RIA, B.R.A.H.M.S. AG, Hennigsdorf/Berlin, Germany;Yamasa Co., Tokyo, Japan) at three sites; a TRAb luminescence immunoassay (LumitestTRAK human LIA, B.R.A.H.M.S. AG, Hennigsdorf/Berlin, Germany) at three sites and twoTRAb enzyme immunoassays (TRAb ELISA, EUROIMMUN Medizinische LabordiagnostikaAG, Lübeck, Germany; Medizym T.R.A., Medipan GmbH, Berlin, Germany) at two sites. The
compared TRAb immunoassays named second generation assays by the manufacturer use
human recombinant or porcine TSHR coated on tubes or microplate wells, respectively.These assays are based on the ability of TRAb to inhibit TSHR binding by labelled bovineTSH. Additionally, the Elecsys
®Anti-TSHR immunoassay was compared with a manual
M22 antibody based TRAb ELISA (Cosmic Ltd, Tokyo, Japan) at one site. This assay usesporcine TSHR coated microplate wells. At least 100 unselected human serum samples wereanalyzed at each site in single determinations spread over at least five runs.
A lot to lot comparison (MP lot vs. P2/P3 lots) of the Elecsys
®Anti-TSHR immunoassay
was performed at three evaluation sites on MODULAR ANALYTICS®E170 and Elecsys®
2010. At least TRAb levels of 50 unselected human serum samples were analyzed in singledeterminations on at least two runs.
All immunoassays were used according to the manufacturer's instructions. Fresh
and frozen serum samples were included in the method comparison study. If frozenserum samples were used, TRAb values of the Elecsys
®assay and the commercial
comparison test were determined from the frozen sample material. The regression datawere calculated according to the method of Passing and Bablok [21].
2.5. Analytical range
The analytical sensitivity was determined in four clinical laboratories using an
analyte-free sample material (Roche Diagnostics) analyzed 21 times in a single run onMODULAR ANALYTICS
®E170 and Elecsys®2010. Mean values and SD of the photon
counts were calculated. The analytical sensitivity was de fined as the mean value plus 2
Table 1
Summary of imprecision study results using control sera (PC THYRO 1 –2) and self-
prepared human pool sera
Materials Range/IU/L Number CV min CVmax CVmedian
Intra-assay imprecision, n=21 replicates in a single run
PC Thyro 1 3.86 –5.80 8 2.1 4.6 2.9
PC Thyro 2 15.28 –22.92 8 0.9 2.4 1.4
Serum pools 0.4 –1.0 8 6.6 21.5 14.9
Serum pools 1.0 –1.5 8 6.0 17.0 9.5
Serum pools 1.5 –2.0 5 4.2 13.9 7.6
Serum pools N2 20 0.8 9.9 3.3
Total imprecision, n=60 replicates in 20 runs
PC Thyro 1 3.86 –5.80 6 4.3 10.1 5.7
PC Thyro 2 15.28 –22.92 6 1.9 4.7 2.4
Serum pools 0.4 –1.0 8 18.1 32.2 28.8
Serum pools 1.0 –1.5 4 13.9 19.9 15.8
Serum pools 1.5 –2.0 4 9.8 15.0 11.9
Serum pools N2 20 1.9 12.4 5.385 D. Hermsen et al. / Clinica Chimica Acta 401 (2009) 84 –89

SD of these measurements. The results were transformed into TRAb values (IU/L) via
calibration curve. Mean values plus 2 SD were calculated from each participatinglaboratory. Finally, a median value of all study results was determined.
The dilution linearity of the Elecsys
®Anti-TSHR immunoassay was evaluated in
three laboratories. Three different human serum samples containing TRAb levels up to
28.0 IU/L were diluted using a self-prepared human low-level ( ≤0.8 IU/L) serum pool.
Each dilution series contained 9 linear dilution steps (9+1, 8+2, 7+3, 6+4, …etc.). In
each run, measurements of the dilutions were performed in duplicate, the undilutedserum and the low-level pool were analyzed in six replicate measurements. Themethod was considered to be linear, if the measured TRAb values differed by less than15% from the expected TRAb concentrations.
2.6. Evaluation of the decision threshold level (cut off limit)
The evaluation of the optimal decision thres hold level for positivity was carried out at
two evaluation sites on MODULAR ANALYTICS
®E170 and Elecsys®2010 systems. TRAb levels
of different collectives were measured. For determination of speci ficity 436 apparently
healthy adults and 210 patients with thyroid disorders others than GD were included in the
study. The cohort of patients with thyroid disorders others than GD was composed of
patients with diagnosis of subacute thyroiditis ( n=91), adenomatousgoiter (45), Hashimotos
disease (27), painless thyroiditis (32), autono mously functioning thy roid nodules (7), toxic
multinodular goiter (1), and 7 other diagnoses. The sensitivity cohort consisted of 102patients with untreated GD. To obtain the optimal cut off limit for positivity a ROC analysiswas performed (20). Sensitivity/speci ficity pairs were calculated by varying the decision
threshold level over the entire measuring range of the TRAb assay. Sensitivity (true positiveresults) was calculated from patients with GD. Speci ficity (true negative results) was
determined from the healthy controls and patients with thyroid disorders others than GD.
3. Results
3.1. Imprecision
Results of the imprecision study are summarized in Table 1 . Median
CVs of provided control sera ranged between 1.4% –2.9% (intra-assay
imprecision) and 2.4% –5.7% (total imprecision), respectively. Regardingthe data of the self-prepared pool sera, median intra-assay CVs were
obtained between 3.3% (TRAb N2 IU/L) and 14.9% (TRAb 0.4 –1.0 IU/L), the
total imprecision study yielded CVs between 5.3% (TRAb N2 IU/L) and
28.8% (TRAb 0.4 –1.0 IU/L). The median total imprecision CV at the
proposed decision threshold level for positivity (TRAb 1.75±0.25 IU/L)
was obtained to be 11.9%.
The FS was assessed on the basis of the total imprecision results
from all evaluation sites. CVs were plotted vs. the means of the
measured control and self-prepared pool sera, respectively. Results
are shown in Fig. 1 . Using the 20% total CV criteria the FS was
determined to be at 0.73 IU/L. As expected the CV at the discussed
c u to f fl i m i t( T R A b1 . 7 5I U / L )w a sf o u n dt ob e1 1 %b yr e g r e s s i o n
analysis.
3.1.1. Inter-laboratory survey
To ascertain that all participating laboratories obtain comparable
TRAb values on MODULAR ANALYTICS®E 170 and Elecsys®2010
systems, an Inter-Laboratory Survey was performed using five different
lyophilized pool sera (samples A –E, target range 1.3 –19.1 IU/L) provided
by Roche Diagnostics. The median values of eight participating
laboratories ranged within the 2 SD limits of each serum pool
(Supplemental Fig. 1A –B), con firming that there was no signi ficant
difference between MODULAR E 170 and Elecsys®2010 test results using
natural samples. Additionally, similar CV of both analyzer systems were
obtained from the study. In the target range 1.3 –2.0 IU/L (samples A –B)
CVs ranged between 3.8% and 26.1 using the Modular system and
between 6.4% and 22.0% using the Elecsys system, respectively.
Regarding TRAb target levels between 5.5 and 19.1 IU/L (samples C –E)
calculated CV ranged between 1.4% and 8.1% and between 1.7% and 4.9%
using the Modular and the Elecsys system, respectively.
Fig. 1. Results from the assessment of total imprecision. Mean of TRAb levels was plotted vs. total imprecision CVs. The FS (CV 20%) was determined at 0.73 IU/L . The CV at the
proposed cut-off limit (1.75 IU/L) was found to be 11%.
Table 2Elecsys anti-TSHR method comparison study: Results of the Passing and Bablok regression analysis evaluated at eight different laboratories
Lab. no. analyzer Method (x) Method (y) Analytical range [IU/L] nr Passing/Bablok
slope/intercept
1—E 170 Medipan, Medizym T.R.A. Elecsys anti-TSHR 0.2 –40.0 175 0.922 1.02 0.05
2—E 2010 BRAHMS TRAb human LIA Elecsys anti-TSHR 1.0 –36.0 51 0.978 1.15 −0.69
3—E 170 BRAHMS TRAb human LIA Elecsys anti-TSHR 0.04 –40.0 116 0.897 1.34 0.56
4—E 170 BRAHMS TRAb human RIA Elecsys anti-TSHR 0.8 –39.0 204 0.955 1.09 −0.35
5—E 2010 Euroimmun TRAb ELISA Elecsys anti-TSHR 0.6 –38.0 112 0.941 1.16 −0.74
6—E 170 BRAHMS TRAb human RIA Elecsys anti-TSHR 0.4 –35.0 409 0.951 1.14 0.01
7—E 170 Cosmic TRAb ELISA Elecsys anti-TSHR 0.4 –32 129 0.895 1.48 −0.22
7—E170 BRAHMS TRAb human RIA Elecsys anti-TSHR 1.0 –57.0 170 0.930 1.09 0.10
8—E 170 BRAHMS TRAb human LIA Elecsys anti-TSHR 0.01 –89 149 0.961 0.95 0.3286 D. Hermsen et al. / Clinica Chimica Acta 401 (2009) 84 –89

3.2. Method comparison study
The method comparison study was carried out in eight clinical
laboratories. TRAb results of five established TRAb immunoassays
were compared with the Elecsys®Anti-TSHR immunoassay. The data
of the regression analysis performed according to Passing and Bablok
are shown in Table 2 . Solely, TRAb values within the technical
measuring range and diluted TRAb values were used for the statistical
evaluation. Correlation coef ficients ranged from 0.895 to 0.978 andslopes from 0.95 to 1.48, respectively. Regression data are illustrated in
representative plots ( Fig. 2 A–E), in which TRAb comparison results
were summarized from sites using the same routine test ( Fig. 2 A,
BRAHMS TRAb human RIA, n=3 sites; Fig. 2 B, BRAHMS TRAb human
LIA, n=3 sites). Elecsys®Anti-TSHR test results compared quite well
with both BRAHMS TRAK human immunoassays. Correlation coef fi-
cients of approximately 0.950 were obtained and the deviation fromslope 1.0 was less than 10% in the pooled data of both comparison
studies. The method comparison study with the TRAb ELISA based on
Fig. 2. (A–E): Representative plots of the method comparison study. Fig. 2A and B shows pooled comparison data of three sites.87 D. Hermsen et al. / Clinica Chimica Acta 401 (2009) 84 –89

porcine TSHR and labelled bovine TSH (Euroimmun; Medipan) yielded
acceptable deviations from slope 1.0 and ranged between 2% and 16%
(Fig. 2 C/D). Correlation coef ficients of 0.922 and 0.941 were observed,
respectively. Although the studied M22 antibody based TRAb ELISA
from Cosmic and the Elecsys®Anti-TSHR test use comparable basic
raw materials TRAb values of the M22 ELISA were signi ficantly lower
compared to the Elecsys TRAb results ( Fig. 2 E). The deviation from
slope 1.0 was found to be 48%. Additionally, the regression coef ficient
of 0.895 proved a high scatter of TRAb results.
Furthermore, the orientation of matched TRAb results relative to
their test speci fic cut off for positivity was compared in all performed
method comparisons. Discordant rates were calculated for each
comparison study and summarized in Table 3 . All measured TRAb
values including those below the detection limit were included in the
study. Comparison results from sites using the same routine test were
pooled. Test speci fic cut off values of the compared routine assays
were used as recommended by the manufacturer. The proposed cut
off for the Elecsys®TRAb assay was 1.75 IU/L. The lowest discordant
rate (3.6%) was obtained from the comparison with the BRAHMS
TRAK human RIA. The comparison of Elecsys®and Medizym TRAb
values revealed the maximum discordant rate of 10.3%. Out of all
method comparison studies a median discordant rate of 7.0% was
calculated.
The lot to lot comparison study (MP-lot vs. P2, P3 lot) performed
in three sites yielded excellent regression data across the technical
measuring range in all evaluated reagent lots. Correlation coef fi-
cients ranged between 0.997 and 1.000 and proved a minimal
scatter of results. The deviation from slope 1.0 ranged between 1%
and 3% and con firmed highly similar TRAb values (Supplemental
Table 1).
3.3. Analytical range
The analytical sensitivity of the Elecsys®Anti-TSHR immunoassay
calculated as the mean value plus 2 SD of a within-run imprecision
ranged between 0.24 and 0.42 IU/L in four participating sites. The final
analytical sensitivity of the Elecsys®Anti-TSHR de fined as the median
value of study results was found at 0.33 IU/L.
The dilution linearity of the Elecsys®Anti-TSHR immunoassay was
assessed in three laboratories performing 9 dilution series containing
TRAb levels up to 28.0 IU/L. Using low level human pool sera as
dilution medium, linearity was obtained in 8 of 9 dilution series.
Deviations of all measured TRAb values were less than 15% of the
expected TRAb values. In a single dilution series a slightly higher
deviation of 15.5% was observed in one dilution step without having
an analytical relevance.3.4. Decision threshold level (cut off limit)
To assess the optimal cut off a ROC analysis was performed using
samples from 436 apparently healthy individuals, 210 patients with
thyroid diseases without diagnosis of Graves' disease (speci ficity
cohort), and 102 patients with untreated Graves' disease (sensitivity
cohort). The ROC analysis revealed an area under curve (AUC) of 0.99
(0.95% CI: 0.98 –1.0) indicating a high sensitivity and speci ficity of the
Elecsys®Anti-TSHR immunoassay. The optimal pair of sensitivity (97%)
and speci ficity (99%) was found at a TRAb level of 1.75 IU/L ( Fig. 3 ).
4. Discussion
Until recently, the routine measurement of TSH receptor (TSHR)
autoantibodies (TRAb) had to rely on manual radioimmunoassays,
ELISA or chemiluminiscence immunoassays, respectively. This was
irrespective on whether the assay system was based on porcine or
human TSH-R for TRAb detection [4]. All these assays were based on
the ability of TRAb to inhibit TSHR binding of labelled bovine TSH [13].
Some years ago, a new assay with a labelled mouse thyroid-stimulating
monoclonal antibody has been developed [14].D u r i n gf o l l o w – u p ,aTable 3
Technical method comparison: Discordant rates evaluated by the relative orientation of TRAb results to the test speci fic cut off for positivity
Routine assay Lab. no. n Elecsys TRAb
≤COElecsys TRAb
NCORoutine TRAb
≤CORoutine TRAb
NCOElecsys TRAb bCO
Routine TRAb NCOElecsys TRAb NCO
Routine TRAb bCODiscordant rate %
BRAHMS TRAb human
LIA Cut off: 1.5 IU/L2 382 205 177 221 161 4 20 6.3
38
BRAHMS TRAb humanRIA Cut off: 1.5 IU/L4 3598 3032 566 3092 506 34 94 3.6
67
Medipan, Medizym T.R.A.
Cut off: 1.5 IU/L1 175 85 90 85 90 9 9 10.3
Euroimmun TRAb ELISA
Cut off: 2.0 IU/L5 199 116 83 124 75 3 11 7.0
Cosmic TRAb ELISA
Cut off: 1.0 IU/L7 129 21 108 13 116 9 1 7.8
Median discordant rate 7.0
TRAb results of five different immunoassays were compared with TRAb results of the Elecsys anti-TSHR immunoassay.
Proposed cut off for Elecsys TRAb immunoassay: 1.75 IU/L.CO: test speci fic cut off value.
Fig. 3. Determination of the cut-off limit for positivity of the Elecsys Anti-TSHR assay.
The ROC analysis was performed in 102 patients with untreated Graves' disease
(sensitivity cohort), 210 patients with other thyroid diseases and 436 apparentlyhealthy individuals (speci ficity cohort). The optimal cut off limit was calculated at
1.75 IU/L (sensitivity 97%, speci ficity 99%). AUC 0.99 (95% CI: 0.98 –1.0).88 D. Hermsen et al. / Clinica Chimica Acta 401 (2009) 84 –89

human thyroid stimulating monoclonal autoantibody (M22)-based
ELISA system for TRAb detection was launched [15]. Aim of the present
study was to evaluate the analytical performance of a newly developed,
fully automated assay system with a M22 technology for measurement
of TRAbs [19]. Based on an international multicentre trial with a large
cohort of patients we demonstrate that the implementation of a TRAb
immunoassay onto fully automated routine analyzers was successful.The evaluation at 9 clinical laboratories generated excellent technical
performance data and demonstrated a high degree of comparability to
established TRAb immunoassays.
With regard to the imprecision study intra-assay and total imprecision
CVs above 1.5 IU/L were found below 10% and 16%, respectively. More
importantly, at the threshold for TRAb positivity (1.75 IU/L, sensitivity 97%,
specificity 99%) the total imprecision CV is found at 11%, approximately.
Our results demonstrate that the Elecsys
®assay meets the general
acceptance criteria for CVs less than 20% close to the cut off for positivity.
These data are somehow in contrast to a previously performed
evaluation of a manual TRAb ELISA with M22-technology which has
been performed at one of the evaluation sites of the present study [18].
At that time the manual assay revealed an insuf ficient calculated
between-run coef ficient of variation of 28% at the recommended cut-
off for TRAb positivity (1.0 IU/L). The data described here support these
findings as the Cosmic ELISA with M22-technology shows the lowest
correlation and the highest deviation from slope 1.0 compared to all
other manual assays ( Fig. 3 e). This discrepancy might be explained by
the different composition of both assays. The manual assay is based on
a solid phase technology using a monoclonal antibody which binds
the TSHR to the surface of the ELISA plate. In contrast, the automated
TRAb assay described here is based on a liquid phase for TSHR-TRAb-
interaction. Therefore, it might be postulated that binding of TRAb to
the relevant epitopes of the TSHR is more stable within the liquid
phase compared to the solid phase [5,13,22 –24]. Moreover, the serum
of patients with GD contains a heterogeneous mixture of TRAbs,
which might comprise antibodies with different functional activities
in addition to “irrelevant ”antibodies. These irrelevant antibodies can
also interfere in certain assays or experimental settings [25,26] which
might also explain the different performances of both M22 assays.
The drawback of all manual assay systems independently of their
individual sensitivity and speci ficity is, however, their labor-intensive
and time consuming assay procedure taking usually at least about 2 to3 h. The great advantage of the fully automated TRAb detection system
described here is the short performance time of less than 30 minwithout
the necessity of collecting a certain number of patient samples as
advisable for manual assays. Another advantage is that testing with the
automated TRAb assay can be integrated into the work flow on routine
laboratory analyzers without splitting of patient samples.
In summary the evaluation of the first fully automated TRAb assay at 9
evaluation sites generated homogeneous performance data that were at
least comparable or superior to existing routine procedures. The test results
in patient samples were well comparable to those generated by established
routine procedures. The data of our multicentre study are in agreement
with recent studies which found no clinically relevant differences [27]or
only slight differences [28] between porcine and human TRAb assay
measurements. Moreover, this assay is less time consuming and needs less
man power than manual TRAb assays and may, therefore, represent the
future technology for TRAb detection in clinical practice.
Acknowledgment
This multicenter study was supported and coordinated by Roche
Diagnostics providing the analyzers and reagents used in the study.
Appendix A. Supplementary data
Supplementary data associated with this article can be found, in
the online version, at doi:10.1016/j.cca.2008.11.025 .References
[1] Weetman AP. Graves' disease. N Engl J Med 2000;343:1236 –48.
[2] Pedersen IB, Knudsen N, Perrild H, Ovesen L, Laurberg P. TSH-receptor antibody
measurement for differentiation of hyperthyroidism into Graves' disease and
multinodular toxic goitre: a comparison of two competitive binding assays. ClinEndocrinol (Oxf) 2001;55:381 –90.
[3] Paunkovic J, Paunkovic N. Does autoantibody-negative Graves' disease exist? A
second evaluation of the clinical diagnosis. Horm Metab Res 2006;38:53 –6.
[4] Schott M, Scherbaum WA, Morgenthaler NG. Thyrotropin receptor autoantibodies
in Graves' disease. Trends Endocrinol Metab 2005;16:243 –8.
[5] Smith BR, Sanders J, Furmaniak J. TSH receptor antibodies. Thyroid 2007;17:923 –38.
[6] Luton D, Le GI, Vuillard E, et al. Management of Graves' disease during pregnancy: the
key role of fetal thyroid gland monitoring. J Clin Endocrinol Metab 2005;90:6093 –8.
[7] Abalovich M, Amino N, Barbour LA, et al. Management of thyroid dysfunction
during pregnancy and postpartum: An Endocrine Society Clinical Practice
Guideline. J Clin Endocrinol Metab 2007;92:S1 –47.
[8] Chan GW, Mandel SJ. Therapy insight: management of Graves' disease during
pregnancy. Nat Clin Pract Endocrinol Metab 2007;3:470 –8.
[9] Schott M, Morgenthaler NG, Fritzen R, et al. Levels of autoantibodies against
human TSH receptor predict relapse of hyperthyroidism in Graves' disease. Horm
Metab Res 2004;36:92 –6.
[10] Quadbeck B, Hoermann R, Roggenbuck U, Hahn S, Mann K, Janssen OE. Sensitive
thyrotropin and thyrotropin-receptor antibody determinations one month after
discontinuation of antithyroid drug treatment as predictors of relapse in Graves'disease. Thyroid 2005;15:1047 –54.
[11] Carella C, Mazziotti G, Sorvillo F, et al. Serum thyrotropin receptor antibodies
concentrations in patients with Graves' disease before, at the end of methimazole
treatment, and after drug withdrawal: evidence that the activity of thyrotropin
receptor antibody and/or thyroid response modify during the observation period.Thyroid 2006;16:295 –302.
[12] Eckstein AK, Plicht M, Lax H, et al. Thyrotropin receptor autoantibodies are
independent risk factors for Graves' ophthalmopathy and help to predict severity
and outcome of the disease. J Clin Endocrinol Metab 2006;91:3464 –70.
[13] Morgenthaler NG. New assay systems for thyrotropin receptor antibodies. Curr
Opin Endocrinol Diabetes 1999;6:251 –60.
[14] Sanders J, Jeffreys J, Depraetere H, et al. Thyroid-stimulating monoclonal
antibodies. Thyroid 2002;12:1043 –50.
[15] Smith BR, Bolton J, Young S, et al. A new assay for thyrotropin receptor autoantibodies.
Thyroid 2004;14:830 –5.
[16] Kamijo K. TSH-receptor antibody measurement in patients with various
thyrotoxicosis and Hashimoto's thyroiditis: a comparison of two two-step assays,coated plate ELISA using porcine TSH-receptor and coated tube radioassay using
human recombinant TSH-receptor. Endocr J 2003;50:113 –6.
[17] Kamijo K. TSH-receptor antibodies determined by the first, second and third
generation assays and thyroid-stimulating antibody in pregnant patients with
Graves' disease. Endocr J 2007;54:619 –24.
[18] Liu C, Hermsen D, Domberg J, et al. Comparison of M22-based ELISA and human-TSH-
receptor-based luminescence assay for the measurement of thyrotropin receptorantibodies in patients with thyroid diseases. Horm Metab Res 2008;40:479 –83.
[19] Gassner D, Golla R, Stock W, Yoshimura Noh J, Sanders J, Rees Smith B.
Development of a new automated assay for thyrotropin receptor autoantibodies
on the Elecsys system. Clin Chem 2007;53(6 Suppl):C –135.
[20] Zweig MH, Campbell G. Receiver-operating characteristic (ROC) plots: a funda-
mental evaluation tool in clinical medicine. Clin Chem 1993;39:561 –77.
[21] Passing H, Bablok W. A new biometrical procedure for testing the equality of
measurements from two different analytical methods. Application of linear
regression procedures for method compa rison studies in clinical chemistry,
Part I. J Clin Chem Clin Biochem 1983;21:709 –20.
[22] Chen CR, Pichurin P, Nagayama Y, Latrofa F, Rapoport B, McLachlan SM. The
thyrotropin receptor autoantigen in Graves disease is the culprit as well as the
victim. J Clin Invest 2003;111:1897 –
904.
[23] Nagayama Y, Rapoport B. Thyroid stimulatory autoantibodies in different patients
with autoimmune thyroid disease do not all recognize the same components of
the human thyrotropin receptor: selective role of receptor amino acids Ser25-
Glu30. J Clin Endocrinol Metab 1992;75:1425 –30.
[24] Sanders J, Oda Y, Roberts S, et al. The interaction of TSH receptor autoantibodies
with 125I-labelled TSH receptor. J Clin Endocrinol Metab 1999;84:3797 –802.
[25] Morgenthaler NG, Tremble J, Huang G, Scherbaum WA, McGregor AM, Banga JP.
Binding of antithyrotropin receptor autoantibodies in Graves' disease serum to
nascent, in vitro translated thyrotropin receptor: ab ility to map epitopes
recognized by antibodies. J Clin Endocrinol Metab 1996;81:700 –6.
[26] Atger M, Misrahi M, Young J, et al. Autoantibodies interacting with puri fied native
thyrotropin receptor. Eur J Biochem 1999;265:1022 –31.
[27] Zophel K, von LP, Roggenbuck D, Wunderlich G, Kotzerke J, Lackner KJ. Are porcine
and human TSH receptor antibody measurements comparable? Clin Lab
2008;54:1 –8.
[28] Hermsen D, Liu C, Domberg J, et al. Comparison of a solid phase human –versus
porcine –thyrotropin receptor-based immunoassay for the measurement of
thyrotropin receptor antibodies in patients with thyroid diseases. Exp Clin
Endocrinol Diabetes 2008;116(Suppl 1):S59 –63.89 D. Hermsen et al. / Clinica Chimica Acta 401 (2009) 84 –89