| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Proteomics and Protein Markers |
1 Department of Pathology, University of Maryland School of Medicine, Baltimore, MD.
2 Department of Laboratory Medicine and Pathology, Hennepin County Medical Center, and University of Minnesota School of Medicine, Minneapolis, MN.
3 Denver Health Medical Center, Denver, CO.
4 Analytical Chemistry Division, National Institute of Standards and Technology, Gaithersburg, MD.
5 Dipartimento di Scienze Cliniche "Luigi Sacco", Università degli Studi di Milano, Milano, Italy.
6 Department of Pathology and Laboratory Medicine, University of California at San Francisco, San Francisco, CA.
7 Departments of Pathology, Cell Biology, Neurobiology and Anatomy, Loyola University Medical Center, Maywood, IL.
8 Member of the American Association for Clinical Chemistry Cardiac Troponin I Standardization Committee.
9 Chair, American Association for Clinical Chemistry Cardiac Troponin I Standardization Committee
aAddress correspondence to this author at: University of Maryland Medical Center, 22 S. Greene Street, Baltimore, MD 21201. Fax 410-328-5880; e-mail rchristenson{at}umm.edu.
 |
Abstract |
|---|
 Top Abstract IntroductionMethods and MaterialsResultsDiscussionReferences |
|---|
20- to 40-fold difference between assays, and better standardization and harmonization are needed. Toward this goal, the AACC cTnI Standardization Committee collaborated with the National Institute of Standards and Technology (NIST) in an earlier study to select 2 candidate reference materials (cRMs). Methods: Two troponin cRMs, a troponin C-troponin I-troponin T (CIT) complex from human heart tissue and a CIT complex from recombinant technology, were supplied to NIST for assessment of composition and purity, and cTnI value assignment. These cRMs and 6 cTnI-positive human serum pools were shipped to manufacturers of 15 cTnI assays. Commutability of the materials was examined by determining the numerical relationship for the cRM preparations between each manufacturer-specified field method and each of the other 14 field methods. These relationships were then compared with the corresponding numerical relationships for the human serum pools. Harmonization of methods was accomplished by determining regression parameters relative to the analytical system yielding values closest to the median for each serum pool. These regression parameters were used to recalculate pool values to harmonize the assays. Interassay CVs before and after harmonization were determined.
Results: Characterization of the CIT and CI cRMs showed that these materials were of specified composition. The proportion of cTnI methods that demonstrated commutability for the CIT cRM was 45%; for the CI cRM, 39% of methods demonstrated commutability. Interassay cTnI variability for the field methods ranged from 82% to 97%, median 88%. After harmonization, variability of the serum pools for the cTnI methods was decreased to between 9.0% and 23%, median 15.5%.
Conclusions: The proportion of methods demonstrating commutability was too low for use as a common calibrator for the cTnI field methods. However a simple strategy using serum pools can improve harmonization of field cTnI methods by more than 5-fold. The CIT cRM was selected by the AACC cTnI standardization committee, and a new lot has been classified as the cTnI certified reference material Standard Reference Material 2921 by NIST.
|
Introduction |
|---|
|
TopAbstractIntroductionMethods and MaterialsResultsDiscussionReferences |
|---|
A single manufacturer has produced all cTnT assays, and measurements for this biomarker are in harmony. On the other hand, there are numerous cTnI measurement systems, all immunoassay based, that have been developed and commercialized by a variety of manufacturers. The nature of cTnI complicates analytical (immunoassay) measurement because the molecule may be found as binary and ternary complexes and free, oxidized, reduced, and phosphorylated forms(9)(10). Confusion regarding cTnI measurements has resulted from the 20- to 40-fold variability demonstrated among the cTnI assays(11)(12). The AACC formed a cTnI Standardization Committee in cooperation with the National Institute of Standards and Technology (NIST) to identify a candidate reference material (cRM) and to decrease intermethod variability among cTnI assays. In an earlier study, the committee used rigorous criteria to narrow the number of cRMs from 10 to 2(12). However, reference materials intended for direct value assignment to manufacturer calibrators should be extensively investigated for commutability(13)(14). The objectives of this study were to characterize the remaining 2 cRMs and determine if 1 could be used to improve cTnI assays. We also aimed to develop a strategy for improving the harmony among cTnI assays. Toward this effort, we examined commutability of the cRMs and investigated how to improve the intermethod variability among cTnI assay systems.
|
Methods and Materials |
|---|
|
TopAbstractIntroductionMethods and MaterialsResultsDiscussionReferences |
|---|
Randomly selected vials of cRMCIT and cRMCI materials were thawed and characterized at NIST for composition and purity by mass spectrometry, liquid chromatography, and amino acid analysis. Estimates of cTnI concentrations for cRMCIT and cRMCI were also determined by mass spectrometry, liquid chromatography, and amino acid analysis. The same methodology was used to quantify both cRMs and is presented in detail for cRMCIT in a recent publication(15). The uncertainty (U) of the measured cTnI concentrations was calculated according to U = kuc; uc is the mean uncertainty in the troponin concentration of the cRM, and the coverage factor, k, was determined from the Student t-distribution corresponding to the appropriate degrees of freedom and represents the 95% confidence interval.
CTNI immunoassay methods
The cTnI systems and manufacturers that participated in this study are listed in Table 1
. The term "field method" refers to the use of manufacturers’ specified methods in accordance with their protocols and calibrator materials.
|
Serum pools
Six human serum pools were prepared from specimens left over after routine analysis at the University of Maryland School of Medicine; appropriate Institutional Review Board approval was obtained. Three of the serum pools, hereafter referred to as SP1, SP2, and SP3, were collected from MI patients between 4 h and 24 h after symptom onset. The cTnI target concentrations for SP1, SP2, and SP3 were 0.4, 1.5, and 15 µg/L, respectively, based on measurements with the Dimension RxL because of routine availability at the University of Maryland. The remaining 3 serum pools, hereafter termed SP4, SP5, and SP6, were collected from MI patients between 48 h and 72 h after symptom onset. The respective Dimension RxL target values were 0.4, 1.5, and 15 µg/L. All samples were collected in phlebotomy tubes containing no anticoagulant, delivered to the laboratory within 1 h, centrifuged at 2000g
for 10 min, and frozen within 4 h at –70 °C. Approximately 30 days of sample collection was necessary to accumulate sufficient volumes for the pools. At the time of pooling, the samples were thawed in a room temperature water bath and the pools prepared within 3 h. Aliquots of 0.8 mL each for the 6 serum pools were pipetted into 1.5-mL screw-top polypropylene low-temperature freezer vials and stored at –70 °C.
We examined the intermethod variability for the relative reactivity of the early (<24 h) vs the late (>48 h) specimens because immunoreactive troponin complexes may show temporal differences after myocardial injury(10)(15). Variability of the relative response of cTnI systems was compared between serum pools of different concentrations by the F-test. P values of <0.05 were considered significant.
A cTnI-negative human serum pool was prepared from 30 apparently healthy volunteers. The cTnI concentration of this pool was below the Stratus CS detection limit of 0.01 µg/L (n = 3 measurements). The Stratus CS was used for this purpose because it was among methods that demonstrated high sensitivity at low cTnI concentrations(11). Twenty-milliliter aliquots of the cTnI-negative pool were prepared for use in the protocol below and stored at –70 °C.
measurement protocol
Vials of cRMCIT and cRMCI, serum pools SP1 through SP6, and cTnI-negative serum were shipped on dry ice to participating manufacturers and maintained at –70 °C upon receipt. According to protocol, just before use the vials of cRMCIT and cRMCI, serum pools SP1-SP6, and the cTnI-negative serum were thawed in a water bath at room temperature. All samples were then vortex-mixed, centrifuged at 2000g for 10 min, and the supernatant used for experimentation. An initial 1:50 dilution of the cRMCIT and cRMCI materials was performed using each respective manufacturer’s recommended diluent. A subsequent 1:20 dilution was performed on each diluted cRM material to produce solutions having a target cTnI concentration of 25 µg/L, using the provided cTnI-negative serum. Further dilution with the cTnI-negative serum produced sets of solutions having nominal cTnI concentrations of 10 µg/L, 5 µg/L, 1 µg/L, and 0 µg/L (i.e., undiluted cTnI-negative serum).
commutability assessment of crm with field methods
Commutability (of a material) has been defined as the ability of a material to yield the same numerical relationships between results of measurements by a given set of measurement procedures, purporting to measure the same quantity, as the relationships obtained when the same procedures are applied to other relevant types of material.2
In this study, commutability for each pairing of cTnI assays was evaluated by assessing how much the measured values for each cRM deviated from the regression relationships defined by analysis of the serum pools SP1 though SP6 by the pair of methods. The assessment of commutability across all possible method pairings was expressed as the "degree of commutability", which is the overall proportion of cTnI assay pairs that showed commutability.
To assess the degree of commutability, SP1 through SP6 and cRMCIT and cRMCI samples in cTnI-negative serum having 3 nominal cTnI concentrations of 1, 5, and 10 µg/L were measured with the field methods by use of the same calibration and in the same run. Measurements were performed in triplicate, and the mean was used for all calculations; all data were sent to the committee for analysis. Because there is no reference method for cTnI measurement, the numerical relationship between measurements of the serum pools and the cRMs were examined for each individual field method compared with each of the other 14 field methods using nonparametric linear regression analysis with the Passing and Bablok method(16). Relative residuals and the standard residual error (Srr) were calculated for the serum pools for each method pair according to the following: method ‘Y’ estimate = (slope *‘X’ comparison value) + y-intercept; relative Residual = | [(method ‘Y’ estimate) – (method ‘Y’ field measurement)] | ÷ (method ‘Y’ estimate);
 |
where n = 6. Commutability of each cRM concentration for each method pair was assessed by calculating its Y estimate with the regression parameters determined for that field method pair. A relative residual for each cRM level was determined according to: relative residual = | [(method ‘Y’ estimate for cRM) – (method ‘Y’ field measurement for cRM)] | ÷ (method ‘Y’ estimate for cRM). Commutability was demonstrated for the nominal cRM level if relative residual was 
3-fold the Srr, hereafter termed the "3Srr criterion", established with serum pools for the method pair; this threshold was adopted from previous work(17) based on the publication of Franzini(18). The cRM for each pair of field method comparisons was commutable if all 3 cRM concentrations were within the 3Srr criterion for the field method pair. As stated above, the degree of commutability for the cRMs was expressed as the overall proportion (percentage) of field method comparisons that demonstrated values within the 3Srr criterion.
An example calculation of the relative residual and the parameters resulting from Passing and Bablok regression analysis are available in the Supplemental Data that accompanies the online version of this article at http://www.clinchem.org/content/vol52/issue09. It must be noted that clinical assays typically have CVs that are large at low concentrations and relatively smaller at high concentrations. However, the Srr is assessed at the average concentration of the samples tested and therefore, depending on the method, the Srr may not accurately reflect variability at lower or higher cTnI concentrations."
harmonization assessment
Harmonization was considered the process for minimizing variability among the cTnI measurement systems. Of the 15 methods used in the study, the field method yielding cTnI values for the serum pools that was closest to the median was designated as the common-comparison system. The common comparison system was used for alignment of the methods by serving as the abscissa in nonparametric linear regression analysis with the Passing and Bablok method(16) for regressing the SP1 through SP6 results with each of the other cTnI systems, which served as respective ordinate values.
The degree of harmonization, indicated by the overall CV among the methods for SP1 through SP6, was determined after measurement with the field methods and for the field methods after compensating for calibration differences with regression parameters from the common-comparison strategy.
The serum pools SP1 through SP6 were also measured after all methods were recalibrated using the cRM preparations as calibrator to test the feasibility or ability of either cRM to serve as a common calibrator for a variety of cTnI assays.
|
Results |
|---|
|
TopAbstractIntroductionMethods and MaterialsResultsDiscussionReferences |
|---|
The degree of commutability according to the 3Srr criterion is summarized for cRMCIT and cRMCI in Table 2
, parts A and B, respectively. For cRMCIT, the degree of commutability was 45% of comparisons within the 3Srr criterion. The degree of commutability for cRMCI was 39% of comparisons meeting the 3Srr criterion.
|
Substantial heterogeneity occurred among the field method comparisons with regard to degree of commutability for the cRMs (Table 2
). For example, data for cTnI methods B, D, and F show a degree of commutability for cRMCIT that was 72% of comparisons meeting the 3Srr criterion (Table 2A
). On the other hand, Table 2A
shows that cTnI methods C, L, and N indicate a degree of commutability for cRMCIT that was 4-fold less at 18%. This difference among cTnI methods occurred with measurement of both cRMs and demonstrates the substantial heterogeneity in commutability of the cRMs among the set of cTnI field methods.
Fig. 1
shows cTnI data for the 6 serum pools when measured with the field methods and after recalibration of the methods with cRMCIT and cRMCI. Fig. 1
illustrates that the variability among methods was similar before and after recalibration with the cRMs. Thus the cRMs did not improve the degree of harmonization between cTnI methods, and a different strategy is necessary to improve agreement among cTnI methods.
|
Fig. 2
shows the degree of harmonization among the field methods for the 6 serum pools before and after correction for calibration differences by adjusting results based on the regression relationship to a selected comparison method. The variability (CV) of the serum pools among the field methods showed a median of 90.5% (range: 82–97%) for the field methods. However after alignment with regression parameters the variability improved to a median CV of 15.5% (range: 9.0–23%). Thus the alignment strategy with human serum pools resulted in a 5-fold improvement in intersystem variability. Fig. 2
summarizes the improvement in harmonization for the cTnI results of SP1 through SP6 before and after alignment of the field methods compared with use of the cRM as a common calibrator.
|
CVs indicating variability of the relative response of cTnI results with the field methods for SP1, SP2, and SP3, collected from MI patients <24 h after symptom onset, ranged from 82%–93% before recalculation with parameters from Passing and Bablok regression and 9%–21% afterward. The variability (CV) of SP4, SP5, and SP6, collected >48 h after onset of symptoms, ranged from 90%–97% before recalculation with the regression parameters and 7.0%–28% afterward. There was no significant difference (P >0.16) in intermethod variability between different levels of cTnI in the early vs late samples.
|
Discussion |
|---|
|
TopAbstractIntroductionMethods and MaterialsResultsDiscussionReferences |
|---|
Commutability is a necessary characteristic of a material that is intended for use as a common calibrator for routine methods. Validating commutability of a material is complicated when there is no reference method for direct evaluation of the material, but can be assessed by comparison of the material’s behavior among available measurement procedures with a good number of serum pools(18). In the case of cTnI, the heterogeneous and ill-defined measurand prohibits development of an extensively commutable calibrator for various assays using different antibodies of different binding characteristics and imperfect specificity. Therefore, it is not surprising that there was substantial heterogeneity among the methods with regard to their response to the cRMs. Variability among the cTnI methods precludes reporting a binary outcome of "commutable" or "not commutable" for the material. The number of method combinations for which these CRMs had the same numerical relationships as observed for the native serum pools was only 45% for cRMCIT.
Neither of the cRMs had commutability properties that can be advocated as suitable for use as a common calibrator for standardization or harmonization of all cTnI assays. It is noteworthy that substantially different degrees of commutability were observed among the set of cTnI measurement procedures. The fact that the degree of commutability varied substantially among methods suggests that there is substantial variability in the interaction between cRMs and the assays, probably because the set of measurement procedures are immunoassays using a variety of antibody reagents, assay formats, and calibration factors to harmonize with predicate devices within the critical measurement range. Also, the cTnI complex structures in the cRMs may not closely approximate those in circulating cTnI. The interplay between the cRM-target analyte differences and the heterogeneous immunoassay measurement tools indicates that a strategy that uses a reference material alone will not achieve harmonization among cTnI measurement procedures. This is not surprising given the complicated issues involved. At the molecular level, the structure of the troponin complex can take numerous forms(9)(10)(21). At the analytical level, various antibodies used in the measurement systems recognize different epitopes with varying avidity and cross-reactivity. The search should continue for a reference material; however, identification of a common calibrating material suitable for standardization of all cTnI methods will be challenging because specimen-dependent variations in cTnI values were found for measurements performed on different cTnI systems that could not be accounted for by calibration differences or by time of specimen collection(22)(23). Although identifying a common calibrating material is important, the primary clinical concern is harmonization of results, i.e., consistency and reliability for interpretation of patient measurements and clinical trial results(24).
Harmonization efforts included the commoncomparison method with alignment of the methods, which aimed to improve congruency between cTnI measurements performed in human serum so that results are coincident among various assay methods throughout the measurement range. With the alignment method using Passing and Bablok regression parameters, we were able to adjust calibration of the cTnI assays and decrease intermethod variability of cTnI results by nearly 5-fold, to an intersystem variability CV of 15%. This improvement in harmonization and decreased intersystem variability is similar to that for other cardiac biomarkers such as CK-MB, at 13%(25), and myoglobin, also at 13%(26). Thus a cTnI harmonization program that uses a panel of serum pools represents a potential approach for substantially improving agreement in results among assays.
On the other hand, reference materials have advantages over serum pools for monitoring the stability of cTnI results across time. The work of Tate et al.(22)(23) showed that a standard reference material was inadequate for improving harmonization, and that use of human serum pools could serve as a secondary reference material for cTnI measurement. However use of human serum does not provide value assignment; therefore, a primary reference material that can be reproduced within defined specifications is essential for monitoring the stability and consistency of cTnI values assigned to serum pools across time.
Previous studies indicated that cTnI recently released from cardiac tissue may undergo molecular changes, within the myocyte(21) and/or perhaps in circulation(9)(10). This leads to the notion that antibodies in analytical systems may not equally recognize cTnI forms that are recently released, i.e., <24 h, vs those that have been circulating longer, i.e., >48 h. A similar study that used fresh serum from individual patients found that harmonization of cTnI measurements was independent of sample collection time after acute coronary syndrome(22). In this study, patient serum pools collected within 24 h of symptom onset, vs serum pools collected >48 h after symptom onset, showed no significant intermethod variability in their relative reactivity. Thus the time of sample collection after MI was not important for preparation of serum pools. Consistent with this finding, Lebugger et al.(21) indicated that molecular cTnI changes were intracellular and did not occur in circulation beyond some minor cleavage of the N and C terminus. Based on this information and our data, we believe that performing separate studies for cTnI measurements in early samples and late samples is unnecessary.
In summary, this study represents one step in efforts toward the important goal of standardization and harmonization of cTnI assays. We found substantial heterogeneity of the cTnI measurement procedures and insufficient commutability of the cRMs for use as a common reference material for calibration of routine methods.
cTnI measurements in serum pools after calibration with cRMCIT and cRMCI did not improve harmonization, indicating that a strategy that includes human serum pools for method alignment is necessary to achieve method agreement. With such a strategy, we demonstrated significant improvement in harmonization of the 15 cTnI assay systems, independent of the reference material. Regarding a recommendation for a cTnI reference material, the AACC cTnI standardization committee has endorsed the cRMCIT material for further study. Through NIST a new lot of this troponin CIT material is available for further study as SRM 2921(15).
|
Acknowledgments |
|---|
|
Footnotes |
|---|
1 Nonstandard abbreviations: cTnI, cardiac troponin I; cTnT, cardiac troponin T; CIT, [troponin C-Troponin I-Troponin T] protein complex; CI, [troponin C-Troponin I] protein complex; AACC, American Association for Clinical Chemistry; cRMs, candidate reference materials; NIST, National Institute of Standards and Technology; IFCC, International Federation of Clinical Chemistry; SDI, standard deviation index. 
2 Commutability definition from: In vitro diagnostic medical devices – Measurement of quantiles in biological samples – Metrological traceability of values assigned to calibrators and control materials. ISO, Geneva, Switzerland. www.clsi.org/Template.cfm?Section=NCCLS_Terminology_Database_Results&Template=/es/source/custom/termsall.cfm. ISO 17511:2003.
|
References |
|---|
|
TopAbstractIntroductionMethods and MaterialsResultsDiscussionReferences |
|---|
The following articles in journals at HighWire Press have cited this article:
|
|
 |
|
  M. Bonham and S. Miller Clinical Comparison of 99th Percentile and 10% Coefficient of Variation Cutoff Values for Four Commercially Available Troponin I Assays Lab Med, August 1, 2009; 40(8): 470 - 473. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. H. B. Wu Gender-Specific Decision Limits for Cardiac Troponin for Risk Stratification? Anesth. Analg., March 1, 2009; 108(3): 686 - 688. [Full Text] [PDF]
|
|
|
|
 |
|
 C.M. Cobbaert, C.W. Weykamp, E.C.H.J. Michielsen, H. Baadenhuijsen, and M.P. van Dieijen-Visser Time-Dependent Instability of Cardiac Troponins in Human Plasma Spiked with NIST Reference Material 2921 Clin. Chem., December 1, 2008; 54(12): 2078 - 2079. [Full Text] [PDF]
|
|
|
|
|
|
G. Cattozzo, E. Guerra, F. Ceriotti, C. Franzini, and on behalf of the Enzyme Working Group of the Itali Commutable Calibrator with Value Assigned by the IFCC Reference Procedure to Harmonize Serum Lactate Dehydrogenase Activity Results Measured by 2 Different Methods Clin. Chem., August 1, 2008; 54(8): 1349 - 1355. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 D. C Gaze and P. O Collinson Multiple molecular forms of circulating cardiac troponin: analytical and clinical significance Ann Clin Biochem, July 1, 2008; 45(4): 349 - 355. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 NACB Writing Group Members, F. S. Apple, R. L. Jesse, L. K. Newby, A. H.B. Wu, and R. H. Christenson National Academy of Clinical Biochemistry and IFCC Committee for Standardization of Markers of Cardiac Damage Laboratory Medicine Practice Guidelines: Analytical Issues for Biochemical Markers of Acute Coronary Syndromes Circulation, April 3, 2007; 115(13): e352 - e355. [Full Text] [PDF]
|
|
|
|
|
|
NACB WRITING GROUP MEMBERS, F. S. Apple, R. L. Jesse, L. K. Newby, A. H.B. Wu, R. H. Christenson, NACB COMMITTEE MEMBERS, R. H. Christenson, F. S. Apple, C. P. Cannon, et al. National Academy of Clinical Biochemistry and IFCC Committee for Standardization of Markers of Cardiac Damage Laboratory Medicine Practice Guidelines: Analytical Issues for Biochemical Markers of Acute Coronary Syndromes Clin. Chem., April 1, 2007; 53(4): 547 - 551. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
) for each pool represents cTnI results for each analytical system using the field methods; the center data set (
) represents cTnI results for the systems using cRMCIT prepared in cTnI-negative serum for calibration; and the third data set (
) shows cTnI results for the systems using cRMCI prepared in cTnI-negative serum for calibration.