| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Letters |
1 Kenneth F. Buechler, Author for correspondence. Biosite Diagnostics Incorporated,
11030 Roselle St., San Diego, CA 92121.
2 Kevin K.Nakamura, Biosite Diagnostics Incorporated, 11030 Roselle St., San Diego, CA 92121
aAuthor for correspondence.
To the Editor:
A recent article [Troponin T and I Assays Show Decreased Concentrations in Heparin Plasma Compared with Serum: Lower Recoveries in Early than in Late Phases of Myocardial Injury (Clin Chem 2000;46:817–21)] concluded that "Until such methods [which are resistant to interference by both heparin and EDTA] are available, the sample of choice for cardiac troponin determinations is serum collected in tubes with or without gel, or in thrombin tubes with and without gel". We believe that the conclusion is not supported by the data presented in the article.
It is clear from the data presented in Table 1
of the article that a
bias exists between plasma and serum measurements for the Elecsys 2010
and the Immulite systems. No data in the article, however, suggest that
the measured bias is related to heparin. With specific reference to
Table 1
, the authors state that the heparin concentrations in the tubes
range from 40 to 70 IU/mL. In their experimental design, they added
heparin to serum tubes at concentrations of 50–450 IU/mL. The authors
have erroneously calculated the concentrations of heparin in the plasma
tubes as listed in Table 1
. In fact, the total heparin in the tubes
ranges from 40 to 70 IU to achieve a final concentration of 14–15
IU/mL . In calculating the heparin concentration in the
tube, the manufacturer does not consider that blood consists of red
blood cells, which presumably may not take up heparin. Therefore,
recalculation of the manufacturer’s stated concentration, assuming a
40% hematocrit, gives final heparin concentrations in the plasma of
between 24 and 25 IU/mL. The association made by the authors
comparing troponin concentrations measured in serum with
added heparin (50–450 IU/mL) and in plasma from heparin tubes with
whole blood (24–25 IU/mL) should not be made because the heparin
concentrations are very different and the sample matrix (serum vs
plasma) is different. The authors state that "Therapeutic
concentrations of heparin at AMI [acute myocardial
infarction] and at cardiac surgery have been estimated to [be] 1
IU/mL and 5 IU/mL, respectively" and that "These correspond to
1.5% and 7% of the concentrations in heparin tubes and do probably
not cause significant in vivo losses of cardiac troponins". The
heparin concentrations during cardiac surgery, ignoring hematocrit
because both matrices are blood, can be up to 
35% of the heparin
concentrations in the blood collection tubes (5 IU/mL/14
IU/mL). If one believes that heparin associates with troponin, then in
vivo therapeutic heparin concentration becomes significant with respect
to heparin in blood tubes.
|
Assuming that heparin does interact with one or more troponin forms at concentrations significantly higher than in heparin tubes, the affinity of the association must be very weak, although the affinity appears to be higher for troponin T than for troponin I. This conclusion is supported by the data in the article, which show that a nearly 10-fold increase in heparin concentration (from 50 to 450 IU/mL) changes the plasma/serum ratios for troponin T only from 86–109% to 51–78%. The article also shows that a nearly 5-fold increase in heparin (from 98 to 450 IU/mL) gives plasma/serum ratios for troponin I of 77% and 69%, respectively. This finding is consistent with heparin affecting the troponin T assay and not consistent with heparin affecting the troponin I assay, as there does not appear to be a significant heparin-dose-dependent decrease in the ratio for the troponin I assay. With reference to Fig. 3 in the article, the authors fail to provide a statistical analysis of the data at each time point, which is necessary to determine whether the indicated changes in the plasma/serum ratio are actually significant. At the earliest time points, the concentrations measured will be the lowest with the greatest error. The error in the ratio is derived from the error of the two measurements. In addition, the authors fail to show whether the plasma/serum ratio changes as a function of time at relevant heparin concentrations (24 IU/mL rather than 98 IU/mL). A hypothesis that was not suggested by the authors, which we would like to put forth, is that a matrix effect is being measured, at least for the troponin I assay. Clearly, plasma and serum are very different, and matrix effects in immunoassays are well known.
We agree with the recent National Academy of Clinical Biochemistry recommendations that suggest using plasma or anticoagulated whole blood for the stat analysis of cardiac markers. The calibration of immunoassays must be specific to the sample type when matrix effects are known to exist. Therefore, immunoassays that are recommended by the manufacturer for use with heparin plasma should be calibrated with heparin plasma to minimize the bias.
BD VacutainerTM Systems. Franklin Lakes, NJ: Becton Dickinson and Company, January 2000:1–7 (URL: www.bd.com).
Wu HB, Apple FS, Gibler BW, Jesse RL, Warshaw MM, Valdes R. National Academy of Clinical Biochemistry Standards of Laboratory Practice: recommendations for the use of cardiac markers in coronary artery diseases. Clin Chem 1999;45:1104–21.
Kenneth F. Buechler, Author for correspondence. Biosite Diagnostics Incorporated, 11030 Roselle St., San Diego, CA 92121. The authors of the article cited above respond
Kevin K.Nakamura, Biosite Diagnostics Incorporated, 11030 Roselle St., San Diego, CA 92121, The authors of the article cited above respond
1 Department of Clinical Chemistry,Helsingborgs Lasarett AB, S-251 87 Helsingborg, Sweden;
2 Department of Clinical Chemistry, University Hospital Lund, S-221 85 Lund, Sweden;
3 Department of Clinical Chemistry and Transfusion Medicine, Ryhov, S-551 85 Jönköping, Sweden;
4 Department of Clinical Chemistry, Karolinska Sjukhuset, S-171 76 Stockholm, Sweden;
5 Medizinische Klinik II, Medizinische Universität zu Lübeck, D-23528 Lübeck, Germany
aAuthor for correspondence.
We have the following replies to the allegations of Buechler and Nakamura (1), from Biosite Diagnostics Inc., of errors in our report, "Troponin T and I Assays Show Decreased Concentrations in Heparin Plasma Compared with Serum: Lower Recoveries in Early than in Late Phases of Myocardial Injury" (2).
First, however, we would like to make a general comment. Obviously, losses of troponin in heparin plasma depend on (at least) three factors: (a) heparin concentration; (b) troponin assay methodology, especially antibodies used; and (c) distribution of troponin forms in patients plasma. The intention of our short report was not to completely elucidate the mechanism by which these factors may interact. Below are our detailed responses to Buechler and Nakamura (1) (quotes in italics).
1. "No data in the article, however, suggest that the measured bias is related to heparin."
"The authors have erroneously calculated the
concentrations of heparin in the plasma tubes as listed in Table 1
. In
fact, the total heparin in the tubes ranges from 40 to 70 IU to achieve
a final concentration of 14–15 IU/mL. In calculating the heparin
concentration in the tube, the manufacturer does not consider that
blood consists of red blood cells, which presumably may not take up
heparin. Therefore, recalculation of the manufacturer’s stated
concentration, assuming a 40% hematocrit, gives final heparin
concentrations in the plasma between 24 and 25 IU/mL."
We, not Becton Dickinson (BD), estimated plasma
concentrations in the BD tubes used (our Table 1
) based on the current
European BD catalog (printed in 2000), and personal information
from BD representatives. However, a statement by a European BD
representative, that "all glass tubes contain a minimum of 143 IU"
was later withdrawn, and in recent discussions with a BD representative
from the United States, we were informed that our European catalog
contains a misprint [sic]. Accordingly, BD tube 367793 should contain
90, not 143 IU per tube.
Neither the BD online catalog nor the current European BD
print catalog provides any estimates of heparin concentrations in blood
or plasma in filled tubes, only amounts of heparin in the respective
tubes. Furthermore, the BD online catalog quoted by Buechler and
Nakamura does not include the catalog numbers of the "European"
tubes used by us, [sic] but different ones, most of which contain
lower amounts of heparin. Table 1
below, based on the European BD print catalog, shows that
heparin amounts in some tubes vary from 72 to 108 IU in plastic and up
to 143 IU in glass tubes. For simplicity, we calculated plasma
concentrations for a hematocrit of 0.5. As long as we do not know the
actual content of heparin in each type of sampling tube, only the
minimum content, this must be a fair approximation. We now understand
from the responses of Buechler and Nakamura and others to our report
(2) that there seems to be rather great confusion around
tube heparin concentrations. Accordingly, in Table 1
we have compared
the values from various international recommendations and the most
recent information on the BD tubes used in our study. We want to
emphasize that the estimates concern minimum heparin concentrations.
Incomplete filling of, e.g., BD tube 367793 containing 143 IU to only 3
mL of whole blood will produce a heparin concentration of 60 IU/mL of
plasma, or 60% of that used in our addition experiments. The
current confusion around actual tube heparin contents confirms the
importance of a point we made in our discussion (2): Sample
tubes for troponin determinations should be validated and specified by
manufacturer and catalog numbers in assay inserts and studies.
3. "The heparin concentrations during cardiac
surgery, ignoring hematocrit because both matrices are blood, can be up
to 35% of the heparin concentrations in the blood collection tubes
(5 IU/mL/14 IU/mL)."
In view of the above, our comparisons between in vitro tube
(Table 1
) and in vivo heparin concentrations are reasonably correct.
4. "Assuming that heparin does interact with one or more troponin forms at concentrations significantly higher than in heparin tubes, the affinity of the association must be very weak ... a nearly 10-fold increase in heparin concentration (from 50 to 450 IU/mL) changes the plasma/serum ratios for troponin T only from 86–109% to 51–78%."
Our study was not designed to estimate molecular affinities, and we do not see any problem in that troponin losses by heparin are not linear. In our view, the observation of a larger decrease of troponin concentrations with an increase of heparin concentration from 50 to 450 IU/mL in two completely different assays justified the suggestion of a binding phenomenon rather than an unspecific matrix effect.
5. "In addition, the authors fail to show whether the plasma/serum ratio changes as a function of time at relevant heparin concentrations (24 IU/mL rather than 98 IU/mL."
If Buechler and Nakamura (1) mean time after sampling, it is correct. We did not even try to answer that question; we believed that troponins are used as stat tests. The data in Table 2 of our article (2) clearly show the change of mean values of P/S ratios ± 95% confidence intervals for three time intervals after onset of chest pain 13–48(1–12, 13–48, and >48 h) calculated from the eight cases shown in the online supplement.
6. "With reference to Fig. 3 in the article, the authors fail to provide a statistical analysis of the data at each time point."
Statistical treatment of serial data for individual patients is an interesting issue (6), but it is far beyond the scope of our short report. A t-test based on a day-to-day imprecision for TnT of 5.6% yields a least significant difference of 16% between two individual measurements at a concentration of 0.1 µg/L. If the imprecision in the measurement of ratio is estimated to 8%, the least significant difference between two samples is 22%. We omitted a discussion of these data in the short report in favor of the group mean values in Table 2.
7. "We agree with the recent National Academy of Clinical Biochemistry Recommendations that suggest using plasma or anticoagulated whole blood for the stat analysis of cardiac markers. The calibration of immunoassays must be specific to the sample type when matrix effects are known to exist. Therefore, immunoassays that are recommended by the manufacturer for use with heparin plasma should be calibrated with heparin plasma to minimize the bias."
We think this approach is highly questionable. An average difference between serum and plasma troponins can easily be calculated, but given the great variability of P/S ratios during myocardial infarction, it seems to be a dangerous oversimplification to use a mean difference as a bias correction.
References
The following articles in journals at HighWire Press have cited this article:
|
|
 |
|
  M. M. Verbeek, H. P.M. de Reus, and C. W. Weykamp Comparison of Methods for the Detection of Oligoclonal IgG Bands in Cerebrospinal Fluid and Serum: Results of the Dutch Quality Control Survey Clin. Chem., September 1, 2002; 48(9): 1578 - 1580. [Full Text] [PDF]
|
|
|
|
|
|
A. Cerutti, L. Corsini, R. Finotto, and C. Perazzi Comparison of Cardiac Troponin I in Serum and Heparin Plasma with the Dimension RxL Assay Clin. Chem., May 1, 2002; 48(5): 790 - 791. [Full Text] [PDF]
|
|
|
|
|
|
K. F. Buechler, K. K. Nakamura, W. Gerhardt, G. Nordin, A. Isaksson, S. Haglund, E. Gustavsson, M. Muller-Bardorf, and H. A. Katus Diagnostic accuracy of an agarose gel electrophoretic method in multiple sclerosis. Clin. Chem., January 1, 2001; 47(1): 144 - 144. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
