| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Letters |
Clinical Chemistry, and Hematology Laboratory, Hospital of Verona, Piazzale A. Stefani 1, 37126 Verona, Italy
aAuthor for correspondence. Fax 39-045-8072156; e-mail romolo.dorizzi{at}mail.azosp.vr.it.
To the Editor:
An editorial in the June issue of the Journal pointed out the usefulness of plasma instead of serum for cardiac troponin I (cTnI) determinations. However, because some analytical systems yield significant differences between serum and plasma troponin concentrations, the author of the editorial advocates more peer-reviewed studies before new assays are implemented in hospital-based laboratories (1). In September 2001 and March 2002, we carried out two comparison studies of cTnI measurements in plasma and serum samples, using different generations of the same assay (Cardiac Troponin-I Flex; Dade Behring). Both serum and plasma samples have been reported as suitable for cTnI determinations in the first (2) and second (3) generations of the assay. We collected two different series of 39 blood samples from patients admitted to the intensive care unit of our hospital during each month. For each patient, a serum (Venoject II, Autosep Gel + Act Z; Terumo Europe) and a plasma specimen (Venoject II, Autosep Gel + lithium heparin, 75 IU/tube) were obtained. cTnI was measured in duplicate in both series with reagents that came with the Cardiac Troponin-I Flex reagent (batch BD1349 in September 2001 and batch CK3031 in March 2002) and cTnI calibrators (batches OMD095 and IHD045, respectively) on the Dimension RxL Analyzer (Dade Behring). Linear regression analysis of the September 2001 results produced a y-intercept of 0.389 µg/L (SE, 0.332 µg/L) and a slope of 0.978 (0.014). The correlation was good [r = 0.9977; 95% confidence interval (CI), 0.9956–0.9988; P <0.0001]. Results for serum [median cTnI value (range), 4.08 (0.30–61.99) µg/L] and plasma [5.50 (0.30–57.58) µg/L] did not show a significant difference (Wilcoxon rank-sum test for paired data).
Linear regression analysis of the March 2002 results yielded a y-intercept of -0.125 µg/L (SE, 0.216) and a slope of 0.941 (0.010). The correlation was good (r = 0.9977; 95% CI, 0.9956–0.9988; P <0.0001). However, results for serum [median cTnI value (range), 5.35 (0.015–77.95) µg/L] and plasma [5.00 (0.01–74.71) µg/L] showed a significant difference (Wilcoxon rank-sum test for paired data, P <0.001). The assay used in September had a mean bias for plasma samples of 1% (95% CI, -2.4% to 4.45%) compared with paired serum samples, whereas the second-generation assay had a mean bias for plasma samples of -11.4% (95% CI, -17.53% to -5.26%). It appears noteworthy that only three pairs of samples had differences >20% (29.5%, 28.3%, and 24.1% at cTnI concentrations of 1.77–4.08 µg/L) in the first-generation assays, whereas seven pairs of samples had differences >20% with the second-generation Dimension assay (68%, 62.5%, 50%, 43.5%, 33.3%, 30.2%, and 26.4%; five at concentrations <0.2 µg/L, one at 0.7 µg/L, and 1 at 
10 µg/L).
Fig. 1
confirms the point made by Dewitte et al. (4) that the uncritical adoption of Bland–Altman plots, far from clarifying, blurs the obtained results. As seen in panels A and B of Fig. 1
, the first-generation, not the revised, Dimension cTnI assay yielded consistent results for serum and plasma. After logarithmic transformation of the x data and construction of a y axis using percentage values (Fig. 1
, C and D), the results obtained for plasma and serum become more consistent. The evaluation steps suggested by international scientific bodies, e.g., IFCC, including anticoagulant validation, should be carried out even when an assay is "simply" reformulated by the manufacturer, and results should be carefully assessed. We agree that caution is required when cTnI results are obtained in different matrices (1)(5); however, heparinized plasma and serum samples appear interchangeable in cTnI measurement with the second-generation Dimension RxL, as can be demonstrated when properly investigated as suggested by Dewitte et al. (4). Finally, even a much smaller group of samples, some of which contained very low concentrations, yielded results similar to those obtained for a much larger group of samples (3)(5).
|
References
The following articles in journals at HighWire Press have cited this article:
|
|
 |
|
  M. Panteghini and F. Pagani On the Comparison of Serum and Plasma Samples in Troponin Assays Clin. Chem., May 1, 2003; 49(5): 835 - 836. [Full Text]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
