Clinical Chemistry
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

Clinical Chemistry 51: 2303-2311, 2005. First published October 13, 2005; 10.1373/clinchem.2005.058180
This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Data Supplement
Right arrow All Versions of this Article:
clinchem.2005.058180v1
51/12/2303    most recent
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Web of Science (2)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Thienpont, L. M.
Right arrow Articles by Tai, S.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Thienpont, L. M.
Right arrow Articles by Tai, S.
Related Collections
Right arrow Endocrinology and Metabolism
(Clinical Chemistry. 2005;51:2303-2311.)
© 2005 American Association for Clinical Chemistry, Inc.

Endocrinology and Metabolism

Feasibility Study of the Use of Frozen Human Sera in Split-Sample Comparison of Immunoassays with Candidate Reference Measurement Procedures for Total Thyroxine and Total Triiodothyronine Measurements

Linda M. Thienpont1,a, Katleen Van Uytfanghe1, John Marriott2, Peter Stokes2, Lothar Siekmann3, Anja Kessler3, David Bunk4 and Susan Tai4

1 Laboratory for Analytical Chemistry, Faculty of Pharmaceutical Sciences, Ghent University, Gent, Belgium.
2 LGC, Teddington, United Kingdom.
3 Institute for Clinical Biochemistry, University of Bonn, Bonn, Germany.
4 National Institute of Standards and Technology, Gaithersburg, MD.

aAddress correspondence to this author at: Laboratory for Analytical Chemistry, Faculty of Pharmaceutical Sciences, Ghent University, Harelbekestraat 72, 9000 Gent, Belgium. Fax 32-9-264-81-98; e-mail linda.thienpont{at}ugent.be.


   Abstract
Top
 Abstract
Introduction
 Materials and Methods
 Results
 Discussion
 References
 
Background: Diagnostic manufacturers must ensure/document metrologically traceable assays. We report on a feasibility study of a split-sample comparison for that purpose. Processed, frozen single-donation sera, assigned target values by candidate reference measurement procedures (cRMPs), were used with immunoassays for total thyroxine (TT4) and triiodothyronine (TT3) as models.

Methods: Two serum panels were quantified for TT4 and TT3 with validated cRMPs and measured in parallel with at least 14 immunoassays. The results were interpreted in terms of traceability of calibration (trueness) and of the individual measurement result (accuracy) by linear regression analysis and graphical representation against specifications. The commutability of the sera was investigated by parallel analysis of TT4 in freshly collected but nonfiltered specimens.

Results: The TT4 (TT3) concentrations in the sera (according to the cRMPs) were 64–269 nmol/L (0.88–13.7 nmol/L). The method comparison showed that for TT4, on average, the immunoassays produced results in agreement with the cRMPs, whereas for TT3, results were typically higher. It also demonstrated a considerable between-assay divergence in traceability of calibration and accuracy. The evidence of noncommutability of the sera attributable to processing, however, indicates that the interpretation should be treated with caution.

Conclusions: Frozen sera can be used for documenting/validating traceability of total thyroid measurements. The way in which the sera are processed may jeopardize commutability, however, and therefore requires in-depth investigation.


   Introduction
Top
 Abstract
 Introduction
Materials and Methods
 Results
 Discussion
 References
 
Trueness-based harmonization of measurement results among laboratories and measurement procedures is a longstanding objective of laboratory medicine [see, for example, Refs. (1)(2)(3)(4)]. In spite of ambitious standardization projects by authoritative organizations such as the International Federation for Clinical Chemistry and Laboratory Medicine and the US Centers for Disease Control and Prevention (5)(6)(7)(8), there are enough reasons to state that even at the beginning of the 21st century standardization has not been realized (9)(10). The concept has received renewed emphasis, however, in light of the recent European Commission Directive on In vitro Diagnostic Medical Devices 98/79/EC (11). Since December 7, 2003, documentation by the clinical diagnostics industry of metrologic traceability of in vitro diagnostic systems has been required. For well-defined analytes, diagnostic systems must produce measurement results in agreement with the true value. Such agreement is determined according to the reference measurement system of the International Organization for Standardization/European Committee for Standardization Standard 17511 (12), which applies a metrologic traceability chain, comprising the SI unit, higher order reference materials, and reference measurement procedures.

We recently established a model reference measurement system to provide the trueness basis for total thyroxine (TT4) 1 and total triiodothyronine (TT3) measurements in serum (13). This work, done in the framework of a project funded by the European Commission (14), made available primary calibrators for T4 and T3 and validated isotope-dilution liquid chromatography–mass spectrometry (ID-LC/MS) candidate reference measurement procedures (cRMPs) and serum-matrix–based reference materials, i.e., 2 panels of 33 frozen sera assigned with TT4 and TT3 target values by the respective cRMPs. Going beyond the project’s objectives, we organized a split-sample comparison, a method comparison based on parallel measurement of the serum-matrix–based reference materials by field immunoassays and cRMPs [see, for example, Refs. (13)(14)(15)] because we considered it necessary to demonstrate that the model reference measurement system effectively facilitates documentation/validation of traceability of field assays. In this context, diagnostics manufacturers worldwide participated with their TT4 and TT3 diagnostic systems.

Here we report on the TT4 and TT3 split-sample comparison, which had 2 principal objectives. The first was to investigate whether sera processed according to a state-of-the-art protocol for steroid hormone reference materials (16)(17) could also be used to prepare reference materials for thyroid hormone measurements. This objective originated from the general requirement for commutability of matrix reference materials (18) and was driven in particular by the awareness that, in thyroid function testing, the equilibrium between protein-bound (99.97%) and free thyroid hormones (0.03%) should be preserved. The second objective was to demonstrate that from a method comparison with split-sample measurement it is possible to document/validate the traceability of both the values assigned to calibrators (trueness) and the individual measurement results (accuracy).


   Materials and Methods
Top
 Abstract
 Introduction
 Materials and Methods
Results
 Discussion
 References
 
reference serum panels
In the split-sample comparison, we used 1 reference serum panel for TT4 and 1 for TT3. The panels, each containing 33 frozen off-the-clot sera obtained from single blood donations from apparently healthy male or female donors, were purchased from Scantibodies Laboratory, Inc. The blood collections were carried out according to accepted protocols for blood banks regulated by the US Food and Drug Administration. One hour after collection and coagulation of the blood at room temperature, serum was isolated by centrifugation. Only those units found negative for the presence of antibodies to HIV I/II and syphilis and for hepatitis B surface antigen were shipped to the processing unit at Scantibodies (refrigerated at 2–8 °C). Three units were enriched with T4 (Aldrich) and T3 (from Henning) standard materials (both with a purity >99% as stated by the manufacturer) to achieve hyperthyroid concentrations; the remaining 30 units were not enriched and were selected on the basis of their TT4/TT3 concentrations by a field assay to span a physiologic concentration range. No preservatives were added, but the sera were filtered through 0.22 µm hydrophilic polyethersulfone filters (Gelman) to assure sterility and were fractionated in 1-mL portions into polypropylene vials. A total of 150–180 serum aliquots were available per donation. The serum aliquots were immediately stored at –70 °C and shipped to the participating laboratories on dry ice with continued storage at –70 °C.

investigation of the commutability of the processed, frozen human serum specimens
To investigate whether processing had influenced the equilibrium between free and protein-bound thyroid hormone, Ghent University performed free-T4 measurements after equilibrium dialysis and then ID-LC/MS measurement of T4 in the dialysate (method under development) and verified by symmetric dialysis (19). In addition, 10 freshly collected sera, obtained by one of the diagnostics companies (coded A in Table 1 ), were analyzed in parallel by their immunoassay and the cRMP at Ghent University. The native serum specimens were obtained from fresh blood collections, clotted at room temperature for 1 h, and then centrifuged for 10 min to separate the serum from the cells. The specimens were divided into aliquots without sterile filtration and frozen. The analysis of the 10 sera by both the immunoassay and the cRMP was done in parallel with 6 sera from this study to guarantee consistency of performance. The relationship between the results from the immunoassay and the cRMP was evaluated with ordinary least-squares regression (OLR).


View this table:
[in this window]
[in a new window]
  		Table 1. Overview of the outcome of the OLR analysis of the TT4 method-comparison data.1

commercial immunoassays
Ten diagnostic companies participated in the study with 15 (for TT4) and 14 (for TT3) immunoassays. The latter were either RIAs [Immunotech (Beckman-Coulter) and RIA-CT (Biosource)] or automated multianalyte immunoanalyzers [Architect (TT4 only) and AxSYM (Abbott Diagnostic Division); ACS 180, ADVIA IMS, Centaur, and Immuno 1 (Bayer); Dimension RxL (Dade Behring); Immulite 2000 (Diagnostics Product Corporation); Vitros Eci (Ortho Clinical Diagnostics); LAS AutoDELFIA (Perkin-Elmer Life & Analytical Sciences); Elecsys 2010 (Roche), and AIA-pack (Tosoh Bioscience)]; the latter assay was performed under standard and modified conditions of equilibration. The manufacturers were left free to adopt the measurement protocol of their choice. They received free samples but were charged for handling and shipment (on dry ice). All companies had the option of anonymous publication of the method comparison or full disclosure of the results.

CRMPS
The sera were analyzed for their TT4 concentration, with cRMPs performed in 4 candidate reference measurement laboratories (cRMLs): Ghent University, LGC (formerly Laboratory of the Government Chemist), University of Bonn, and NIST. As described elsewhere, all laboratories used a validated variant of an ID-LC/MS cRMP (13). During measurement of the serum panel, 2 sera previously measured by all cRMLs were used for internal precision and accuracy control to verify fulfillment of predefined performance criteria, i.e., maximum within-run and total CVs of 1.5% and 2.0%, respectively, and a maximum systematic deviation of 0.9%. In addition, the cRMLs worked under network conditions and had to achieve maximum between-laboratory CVs of 2.5% (13). All calibration was done with the same material, the candidate T4 certified reference material, made available through this study by the Institute for Reference Materials and Measurements (IRMM) from the Joint Research Centre (Geel, Belgium). The reference material mean [expanded uncertainty (U)] purity was certified to be 98.72 (0.30)% (k = 2). Measurement of the 33 sera was organized in such a way that 3 sets (from 3 cRMLs) of duplicate (measurement of each serum in 2 independent measurement series) results were available per serum. The uncertainty of measurement was calculated in compliance with International Organization for Standardization guidelines (13)(20).

For logistic reasons (shipment on dry ice), the TT4 and TT3 method comparisons were organized at the same time. Because the TT3 network was not yet established, the TT3 panel was analyzed only by Ghent University. The targeting cRML used an ID-LC/tandem MS measurement procedure validated against ID–gas chromatography–MS (21) and identified by the Joint Committee for Traceability in Laboratory Medicine as a higher order cRMP (22). For calibration, a candidate T3 certified reference material from IRMM, with a mean (U) certified purity of 97.17 (0.23)% (k = 2), was used. The sera were measured 3 times as singletons in 3 independent measurement series. For internal precision and accuracy control, 2 lyophilized sera, previously assigned with target values by parallel measurement between Ghent University and NIST (23), were used to verify that the cRMP was performed according to the same performance criteria as specified for TT4 (24). In addition, the expanded uncertainty was estimated for the TT3 measurements. After the method comparison was completed, the University of Bonn finalized the development of its cRMP for TT3 and in the process cross-checked the TT3 measurements by Ghent University.

statistical analysis and graphical presentation
We treated the data of the split-sample comparison statistically with OLR analysis because the comparison method was of higher hierarchical order and a linear relationship was expected. We also present the results graphically in scatter and difference plots. We excluded the results for the enriched samples from the OLR analysis, but because of their importance for validation of calibration, we show the extended regression lines in the scatter plots. We calculated the deviation (%) of the results by the respective field assays from the target values by the ID-LC/MS cRMPs via OLR at the lowest, middle, and highest values. We plotted the residuals (%) vs the ID-LC/MS cRMPs target values. Before doing so, however, we normalized the immunoassay values on the basis of the OLR equations. In the residual plots, we included limits for the allowable total error (TE) of a single measurement result as derived from data on biological variation, 7.0% for TT4 and 12.0% for TT3 (25).


   Results
Top
 Abstract
 Introduction
 Materials and Methods
 Results
Discussion
 References
 
Because only a few companies agreed to disclose their results, the method comparison is presented in an anonymous way. Note, however, that a company that participated for both TT4 and TT3 measurements is designated by the same code.

The endogenous TT4 (TT3) concentrations by the ID-LC/MS cRMPs were 64–136 nmol/L (0.88–2.45 nmol/L); the enriched sera had target concentrations of 224–269 nmol/L (4.8–13.7 nmol/L).

For TT4, the expanded ID-MS measurement uncertainties calculated for each serum separately were 1.6%–3.0% with a mean of 2.2% (13). The uncertainty values differ slightly from those reported by Thienpont et al. (13) because the purities of the IRMM candidate T4 and T3 reference materials have become certified; hence, the associated uncertainties have been taken into account. For the TT3 measurements, the mean estimated expanded uncertainty was 6.0% (k = 4.3 for the measurement protocol by only 1 cRML in triplicate). After the method comparison was completed, however, the reliability of the results obtained by Ghent University was confirmed by blind measurements at the University of Bonn (mean deviation, 1.8%; range, –3.1% to 4.9%). Overviews of the outcomes of the OLR analyses for TT4 and TT3 are given in Tables 1Up and 2 , respectively, including the deviations ({Delta}%) at the lowest, middle, and highest ID-LC/MS target concentrations. In each table the first row represents the OLR data for the mean of the measurement results by all immunoassays compared with the cRMP target values. Scatter plots of the mean of all results compared with the ID-LC/MS cRMPs for TT4 and TT3 are presented in Fig. 1A and Fig. 2A , respectively. Although the values of the enriched sera were omitted from OLR analysis, the extended regression lines are shown. The combined scatter plots for the individual immunoassays are shown in Fig. 1B and Fig. 2B , but in these plots the enriched sera are not shown. The data from Tables 1Up and 2 are shown in graphical form, as scatter and absolute difference plots, in the Data Supplement that accompanies the online version of this article at http://www.clinchem.org/content/vol51/issue12. A selection of residual plots, including TE limits, is shown in Figs. 3 and 4 (for an overview of all residual plots, see the accompanying online Data Supplement). The scatter plot and regression equations of the results by immunoassay A compared with the cRMP for the samples used for commutability testing are shown in Fig. 5 .


View this table:
[in this window]
[in a new window]
  		Table 2. Overview of the outcome of the OLR analysis of the TT3 method-comparison data.1



View larger version (17K):
[in this window]
[in a new window]
  		Figure 1. TT4 method comparison between immunoassays and the cRMPs.

(A), scatter plot with OLR line representing the mean of all immunoassays/cRMPs. Although the enriched sera were not included in the regression analysis, the scatter plot includes their concentrations and shows the extended regression line (dotted line). (B), combined scatter plot (without the enriched sera) for the different method pairs; i.e., TT4 immunoassays A to O/cRMPs.



View larger version (17K):
[in this window]
[in a new window]
  		Figure 2. TT3 method comparison between immunoassays and the cRMP.

(A), scatter plot with OLR line representing the mean of all immunoassays/cRMP. Although the enriched sera were not included in the regression analysis, the scatter plot includes their concentrations and shows the extended regression line (dotted line). (B), combined scatter plot (without the enriched sera) for the different method pairs; i.e., TT3 immunoassays A to O/cRMPs.



View larger version (18K):
[in this window]
[in a new window]
  		Figure 3. Selected residual plots (after normalization of the results) for the TT4 method comparison.

Top plot represents the mean of the results obtained by all immunoassays. Dashed lines indicate limits for allowable TE for a TT4 single measurement (7%).



View larger version (18K):
[in this window]
[in a new window]
  		Figure 4. Representative residual plots (after normalization of the results) for the TT3 method comparison.

Top plot represents the mean of the results by all immunoassays. Dashed lines indicate limits for allowable total error for a TT3 single measurement (12%).



View larger version (17K):
[in this window]
[in a new window]
  		Figure 5. Scatter plot showing the results obtained by the commercial assay that performed differently with the samples provided by the European project ( {permzspch055}) and newly acquired frozen samples in comparison with the cRMP ({cjs2108}).

Regression equation for samples provided by the European project: y = 1.149x + 9.1 nmol/L). Regression equation for the newly acquired frozen samples: y = 1.119x – 3.2 nmol/L.

The free T4 concentrations were 44–133 pmol/L.


   Discussion
Top
 Abstract
 Introduction
 Materials and Methods
 Results
 Discussion
References
 
justification of sample material used
The approach of method comparison with parallel measurement of frozen off-the-clot human sera is generally accepted as the basis for the establishment/documentation/validation of the traceability of field measurement procedures to the SI (26)(27). In this study, in dealing with SI traceability of serum TT4/TT3 measurements, we opted to use sera obtained from single donations with thyroid hormone concentrations within the reference intervals. To also cover the upper measurement ranges of field assays, we enriched 3 sera with T4 and T3. Because we were not certain that these sera and sera with endogenous hormones would respond identically to immunoassays, we excluded the measurement results for enriched sera from the OLR analysis but show them in the graphical plots.

With respect to the quality of matrix reference materials, there are 2 requirements: long-term stability and commutability. The first requirement was not a prime concern in this study because the split-sample comparison could be organized within a maximum of 8 months. Considering, nevertheless, that the project was a feasibility study, we performed, under conditions recommended by IRMM, a short (2 weeks) and a long-term (12 and 18 months) isochronous stability study. IRMM confirmed the stability of the materials and predicted shelf lives until 2008 (TT3) and 2012 (TT4).

The commutability requirement of matrix reference materials is necessary because, although prepared from human sources, processed materials may acquire properties that make them behave differently from native patient samples in immunoassays. In this respect, there is a guideline available from the Clinical and Laboratory Standards Institute (CLSI; formerly the NCCLS) that describes the procedures for preparing frozen human serum pools of high-quality suitable for cholesterol measurement (18). However, in certain aspects, the protocol is so demanding that it almost becomes impractical to perform with a typical collection and processing center. Because of these logistic problems, we decided to prepare the sera for this study according to the state-of-the-art process as used for steroid hormones (16)(17) and to assess whether it would be feasible for application to thyroid hormones. In this respect, we focused on the preservation of the equilibrium between the protein-bound and free thyroid fractions after processing and considered the unexpectedly high free T4 concentrations a potential indication that the commutability of the sera had been compromised. We therefore compared for 1 commercial TT4 immunoassay the behavior of the processed sera with freshly collected specimens. For logistic reasons (shipment to Ghent University), this experiment also necessitated freezing of the fresh sera. However, it was confirmed beforehand that the freezing/thawing process did not influence the measurement results of the immunoassay. Evidence for a different behavior of the 2 types of sera in the measurement of TT4, at least with the immunoassay used, can be seen in Fig. 5Up . Although the cause of the noncommutability has not been determined, we suspect the filtration process because this was the main difference between the 2 types of sera. On the other hand, the problem may also be attributable to the immunoassay, for which the sera seem to have a lack of commutability, because principally, it is not expected that an in vitro imbalance between the free and total T4 fractions in a serum affects the TT4 measurement results. On the basis of these observations, the project leader immediately urged the manufacturers not to recalibrate their immunoassays on the basis of the performed split-sample comparison. We also concluded that for preparation of serum reference materials of appropriate quality suitable for the establishment/validation of traceability of TT4/TT3 measurement, an in-depth study of the protocol to process the materials would be necessary. That study must investigate with several immunoassays, for example, how the filtration process, the time that samples are kept refrigerated before being divided into aliquots and frozen, or the thawing temperature influence the commutability of the reference materials. It is proposed to carry out this study within the International Federation for Clinical Chemistry and Laboratory Medicine framework, i.e., the Working Group for Standardization of Thyroid Function Tests (28).

interpretation of the split-sample comparison
Traceability of calibration (trueness).
Principally, a split-sample comparison study provides an evaluation of the agreement of an assay’s results with those obtained by the cRMP(s). In case of agreement, the traceability of the values assigned to the assay’s calibrators (trueness of performance) is demonstrated. This conclusion can be drawn from OLR analysis of the method-comparison results and requires a slope and intercept not significantly deviating from 1 and 0, respectively. The OLR data tabulated for TT4 in the first row of Table 1Up show that, on average, the TT4 immunoassays were in agreement with the cRMP values (see also Fig. 1AUp ). For the TT3 immunoassays, as can be seen from Table 2Up (and Fig. 2AUp ), they are on average "overcalibrated" (slope >1 and intercept = 0). However, it should be emphasized that these interpretations are correct only provided that the reference materials are commutable with the tested assays. If not, a split-sample comparison does not necessarily reflect the trueness of performance of an immunoassay on native patient sera. Therefore, in view of the evidence of jeopardized commutability in this feasibility study, it may be better to do the interpretation in a relative way. On this basis, the OLR data in Table 1Up (and also Fig. 1BUp ), showing divergence in slopes and intercepts for the individual assays, indicate considerable differences in calibration status among the TT4 immunoassays. From the {Delta}% values in Table 1Up it is obvious that for some TT4 immunoassays the systematic differences are proportional to the concentration, whereas for others, they are rather constant. For the TT3 immunoassays, as can be seen in Table 2Up , between-immunoassay divergence (see also Fig. 2BUp ) was also considerable and systematic differences were proportional to the concentration (see the {Delta}% values). Thus, although the potential lack of commutability prohibits us from concluding which immunoassay has SI-traceable calibration, the observations demonstrate the general need for recalibration of TT4 and TT3 immunoassays. A detailed presentation of these observations is available in the different plots shown in the accompanying online Data Supplement.

With respect to the question of whether enriched sera have utility for judging traceability of calibration, the extension of the regression lines in Fig. 1AUp shows that in TT4 immunoassays, sera 31–33 behaved, on average, as sera with endogenous hormone. On the other hand, for TT3, on average, the results for the enriched sera did not coincide with the extended regression line (Fig. 2AUp ).

Traceability of the individual measurement result (accuracy).
A method comparison against a cRMP with single-donation sera also allows evaluation of the traceability of measurement at the level of the individual serum sample. The accuracy of performance depends mainly on the precision and specificity of the field assay. These characteristics are reflected in a split-sample comparison by the correlation coefficient and residuals, i.e., the higher the precision/specificity, the higher the correlation coefficient and the lower the residuals. In other words, a high variability along the regression line demonstrates a large random error component in the method comparison, attributable to high imprecision, susceptibility of the field assay to sample-related effects, or a combination of both. By sample-related effects, we mean the effects attributable to the individual matrix. For example, in Table 1Up , for TT4 the correlation coefficient for the mean of all results is 0.998, but only 3 assays have a correlation coefficient ≥0.99. For TT3 (Table 2Up ), the correlation coefficient for the mean of all results is 0.986, but at the individual level, the best rank is a correlation coefficient ≥0.975.

With respect to deriving sample-related effects, we made use of the residual plots, after normalization of the values for each immunoassay via the OLR equation. The TE limits in the plots allow direct visual interpretation of acceptable accuracy, i.e., as long as the residuals are within the limits. Note that, strictly speaking, evaluation against a TE criterion supposes a singleton measurement; thus, the manufacturers should interpret the residual plot with caution depending on how the reported measurement results were obtained. Visual inspection of Figs. 3Up and 4Up shows that, on average (first plot), the situation is ideal for TT4 and good for TT3. However, at the individual level, considerable differences among the immunoassays is again apparent.

It is also important to note that the combination of traceability of calibration and of measurement at the level of the individual sample is important. For example, from Table 1Up and Fig. 3Up it can be derived that TT4 assay L shows an acceptable traceability of calibration with a high correlation coefficient and all residuals (%) within the TE limits, which is of course the ideal situation. Other assays, such as TT3 immunoassay C (Table 2Up ), showed acceptable traceability of calibration with a lower correlation coefficients and too many residuals outside the TE limits (for the latter, see the accompanying online Data Supplement). This situation reflects a contribution of the random error component in the immunoassay results that is too high and hence requires improvement of the assay’s precision and/or specificity. On the other hand, some assays (e.g., assay O) combined for both TT4 and TT3 a slope significantly deviating from 1 and a high degree of unreliability [as evidenced by the SE of the slope of, respectively, 14% (T4) and 19% (TT3)], a low correlation coefficient, and too many residuals outside the TE limits (see Figs. 3Up and 4Up ). In that case, it is not worthwhile correcting the calibration because the high random error component in the method comparison would produce an unacceptably high uncertainty. However, because of the evidence for noncommutability of the sera in this feasibility study, we again recommend caution with the interpretation of the documented accuracy.

In conclusion, through this study we showed that panels of frozen single-donation sera, assigned target values by (a) cRMP(s), principally serve the purpose of documentation/validation of traceability of commercial immunoassays for total thyroid hormone measurements via split-sample comparison. Through the practical organization of the method comparison and participation of diagnostics manufacturers all over the world, we demonstrated that from the logistic point of view it is feasible. Although pooled sera are useful for documentation/validation of traceable calibration, the feasibility study showed that the added value of a method comparison with single-donation sera is that the traceability of the individual measurement results (accuracy) can also be judged. However, the correctness of the interpretations depends on the commutability of the reference materials. From this perspective, the presented feasibility study provided evidence that for field thyroid hormone measurements it will be necessary to develop a protocol for processing materials with appropriate intermethod commutability and to perform a final commutability test with all immunoassays as described in the CLSI EP14 protocol (29). Although we restricted our study to documentation/validation of metrologic traceability, it is a logical consequence that a method comparison with a cRMP on commutable serum-matrix reference materials would be an ideal tool for diagnostics manufacturers to establish traceable assays. They could also take advantage of such panels during the development of their assays, e.g., for testing antibody specificity and optimizing the immunoreaction conditions. Last but not least, the voluntary participation of international diagnostics manufacturers showed their interest and willingness to meet both the scientific and legal demand of establishing metrologic traceability of their assays.


   Acknowledgments
 
Certain commercial equipment, instruments, materials, and companies are identified in this report to adequately specify the experimental procedure. Such identification does not imply recommendation or endorsement by the European Commission or the NIST, nor does it imply that the equipment, instruments, materials, and companies identified are the best available for the purpose. The European authors gratefully acknowledge financial support by the European Commission of project G6RD-CT-2001-00587. All of the authors are also grateful for the constructive discussions with the project members from the in vitro diagnostics industry, representing EDMA and AdvaMed. Last but not least, the project work package leader is indebted to the fruitful discussion of the split-sample comparison data with D. Stöckl from STT-Consulting.


   Footnotes
 
1 Nonstandard abbreviations: (T)T4, (total) thyroxine; (T)T3, (total) triiodothyronine; ID-LC/MS, isotope-dilution liquid chromatography–mass spectrometry; cRMP, candidate reference measurement procedure; OLR, ordinary least-squares regression; cRML, candidate reference measurement laboratory; IRMM, Institute for Reference Materials and Measurements; TE, total error; and CLSI, Clinical and Laboratory Standards Institute.


   References
Top
 Abstract
 Introduction
 Materials and Methods
 Results
 Discussion
 References
 

  1. Cali JP. An idea whose time has come. Clin Chem 1973;19:291-293.[Web of Science][Medline] [Order article via Infotrieve]
  2. Uriano GA, Gravatt CC. The role of reference materials and reference methods in chemical analysis. CRC Crit Rev Anal Chem 1977;6:361-411.
  3. Tietz NW. A model for a comprehensive measurement system in clinical chemistry. Clin Chem 1979;25:833-839.[Free Full Text]
  4. Stamm D. Reference materials and reference methods in clinical chemistry. J Clin Chem Clin Biochem 1979;17:283-297.[Web of Science][Medline] [Order article via Infotrieve]
  5. Maas AH, Weisberg HF, Burnett RW, Muller-Plathe O, Wimberley PD, Zijlstra WG, et al. Approved IFCC methods. Reference method (1986) for pH measurement in blood. International Federation of Clinical Chemistry (IFCC), expert panel on pH, blood gases and electrolytes. J Clin Chem Clin Biochem 1987;25:281-289.[Web of Science][Medline] [Order article via Infotrieve]
  6. Marcovina SM, Albers JJ, Dati F, Ledue TB, Ritchie RF. International Federation of Clinical Chemistry standardization project for measurements of apolipoproteins A-I and B. Clin Chem 1991;37:1676-1682.[Abstract/Free Full Text]
  7. Siekmann L, Bonora R, Burtis CA, Ceriotti F, Clerc-Renaud P, Ferard G, et al. IFCC primary reference procedures for the measurement of catalytic activity concentrations of enzymes at 37 degrees C. Part 1. The concept of reference procedures for the measurement of catalytic activity concentrations of enzymes. Clin Chem Lab Med 2002;40:631-634.[CrossRef][Web of Science][Medline] [Order article via Infotrieve]
  8. McNamara JR, Leary ET, Ceriotti F, Boersma-Cobbaert CM, Cole TG, Hassemer DJ, et al. Point: status of lipid and lipoprotein standardization. Clin Chem 1997;43:1306-1310.[Abstract/Free Full Text]
  9. Stenman UH. Immunoassay standardization: is it possible, who is responsible, who is capable? [Opinion]. Clin Chem 2001;47:815-820.[Free Full Text]
  10. Klee GG. Clinical interpretation of reference intervals and reference limits. A plea for assay harmonization [Review]. Clin Chem Lab Med 2004;42:752-757.[CrossRef][Web of Science][Medline] [Order article via Infotrieve]
  11. . . Directive 98/79/EC of the European Parliament and of the Council of 27 October 1998 on in vitro diagnostic medical devices. OJEC 1998; Dec 7;:L 331/1-L 331/37.
  12. . International Organization for Standardization. In vitro diagnostic medical devices. Measurement of quantities in samples of biological origin—metrological traceability of values assigned to calibrators and control materials. EN/ISO 17511 2003 ISO Geneva. .
  13. Thienpont LM, Van Uytfanghe K, Marriot J, Stokes P, Siekmann L, Kessler A, et al. Metrologic traceability of total thyroxine measurements in human serum: efforts to establish a network of reference measurement laboratories. Clin Chem 2005;51:161-168.[Abstract/Free Full Text]
  14. . European Commission—5th Framework Programme. Competitive and Sustainable Growth Project G6RD-CT-2001-00587 (2001–2005). Feasibility studies for the development of reference measurement systems for thyrotropin (TSH) and for free thyroxine (FT4), and validation of reference measurement systems (procedure and material) for thyroxine (T4) and triiodothyronine (T3) in human serum (2001–2005) 2005 Directorate General for Research, European Commission Brussels, Belgium. .
  15. . Clinical and Laboratory Standards Institute. Method comparison and bias estimation using patient samples; approved guideline—second edition 2002 CLSI Wayne, PA. CLSI Document EP9–A2..
  16. Thienpont LM, De Leenheer AP. Efforts by industry toward standardization of serum estradiol-17ß measurements [Technical Brief]. Clin Chem 1998;44:671-674.[Free Full Text]
  17. Thienpont LMR, De Leenheer AP, Siekmann L, Kristiansen N, Linsinger T, Schimmel H, et al. The characterisation of cortisol concentrations in a serum reference panel (IRMM/IFCC No 451). European Commission, Directorate General, Joint Research Centre, IRMM/IFCC Information, Reference Materials, EUR 19007 EN 1999 European Commission, Directorate General, Joint Research Centre Brussels, Belgium. .
  18. . Clinical and Laboratory Standards Institute. Preparation and validation of commutable frozen human serum pools as secondary reference materials for cholesterol measurement procedures; approved guideline 1999 CLSI Wayne, PA. CLSI Document C37-A..
  19. Ross HA, Huebner FR, Benraad TJ, Kloppenborg PWC. Free thyroid hormone assays in undiluted serum by symmetric dialysis. Albertini A Ekins RP eds. Free hormones in blood 1982:113-120 Elsevier Amsterdam. .
  20. . International Organization for Standardization. ISO guide to the expression of uncertainty in measurement (GUM) 1995 ISO Geneva. .
  21. Thienpont LM, Fierens C, De Leenheer AP, Przywara L. Isotope dilution-gas chromatography/mass spectrometry and liquid chromatography/electrospray ionization-tandem mass spectrometry for the determination of triiodo-L-thyronine in serum. Rapid Commun Mass Spectrom 1999;13:1924-1931.[CrossRef][Web of Science][Medline] [Order article via Infotrieve]
  22. . Joint Committee for Traceability in Laboratory Medicine (JCTLM). JCTLM database of reference materials and reference measurement procedures http://www1.bipm.org/en/committees/jc/jctlm/jctlm-db/ (accessed July 2005)..
  23. Tai SS, Bunk DM, White E, 5th, Welch MJ. Development and evaluation of a reference measurement procedure for the determination of total 3,3',5-triiodothyronine in human serum using isotope-dilution liquid chromatography-tandem mass spectrometry. Anal Chem 2004;76:5092-5096.[Medline] [Order article via Infotrieve]
  24. Thienpont LM, Franzini C, Kratochvila J, Middle J, Ricós C, Siekmann L, et al. Analytical quality specifications for reference methods and operating specifications for networks of reference laboratories. Recommendations of the European EQA-Organizers Working Group B [Review]. Eur J Clin Chem Clin Biochem 1995;33:949-957.[Web of Science][Medline] [Order article via Infotrieve]
  25. Desirable specifications for total error, imprecision, and bias, derived from biologic variation http://www.westgard.com/biodatabase1.htm (accessed July 2005)..
  26. Lasky FD. Achieving accuracy for routine clinical chemistry methods by using patient specimen correlations to assign calibrator values. A means of managing matrix effects. Arch Pathol Lab Med 1993;117:412-419.[Web of Science][Medline] [Order article via Infotrieve]
  27. . Clinical and Laboratory Standards Institute. Standardization of sodium and potassium ion-selective electrode systems to the flame photometric reference method; approved standard—second edition 2000 CLSI Wayne, PA. CLSI Document C29–A2..
  28. IFCC. www.ifcc.org/divisions/CPD/handbook/Handbook_2003-2005/ (accessed July 2005)..
  29. . Clinical and Laboratory Standards Institute. Evaluation of matrix effects; proposed guideline—second edition 2005 CLSI Wayne, PA. CLSI Document EP14–A2..



The following articles in journals at HighWire Press have cited this article:


Home page
 Clin. Chem.Home page
H. W. Vesper and L. M. Thienpont
Traceability in Laboratory Medicine
Clin. Chem., June 1, 2009; 55(6): 1067 - 1075.
[Abstract] [Full Text] [PDF]

eLetters:

Read all eLetters

TT3 Values
Dr. Guillermo Acebedo, et al.
Clinical Chemistry Online, 28 Nov 2005 [Full text]
This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Data Supplement
Right arrow All Versions of this Article:
clinchem.2005.058180v1
51/12/2303    most recent
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Web of Science (2)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Thienpont, L. M.
Right arrow Articles by Tai, S.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Thienpont, L. M.
Right arrow Articles by Tai, S.
Related Collections
Right arrow Endocrinology and Metabolism

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS