| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Articles |
a Address for correspondence: Institut für Klinische Chemie und Pathobiochemie, Klinikum rechts der Isar, Ismaninger Strasse 22, D-81675 München, Germany. Fax 49 89 4140 4875; e-mail Steimer{at}mail.KlinChem.med.TU-Muenchen.de
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Methods: In a comprehensive evaluation, CsA was analyzed in 145 samples with the new CEDIA® assay and compared with the Emit® assay with the old and new pretreatments, the TDx® monoclonal and polyclonal assays, the AxSYM®, and HPLC. All samples were from patients with liver and/or kidney transplants.
Results: The CEDIA offered the easiest handling, followed by the AxSYM, which showed the longest calibration stability. The TDx monoclonal assay provided the lowest detection limit and the lowest CVs. The mean differences compared with HPLC were as follows: Emit, 9–12%; CEDIA, 18%; AxSYM, 29%; and TDx monoclonal, 57%. The CycloTrac® RIA paralleled the Emit results. In contrast to the mean differences, substantial (>200%) and variable overestimations of the CsA concentration were observed in individual patient samples. Metabolic ratios, estimates of the overall concentrations of several cross-reacting metabolites (nonspecific TDx polyclonal/specific reference method), correlated with the apparent biases of the various monoclonal assays. Metabolic ratios varied up to 10-fold, which translated into biases for individual samples between -7% and +174%. The higher the cross-reactivity of an assay was, the higher was the range of biases observed. The interindividual differences markedly exceeded other factors of influence (organ transplanted, hepatic function).
Conclusion: Because assay bias cannot be predicted in individual samples, substantially erratic CsA dosing can result. The specificity of CsA assays for parent CsA remains a major concern. © 1999 American Association for Clinical Chemistry
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Considerably different biases have been published and conflicting recommendations have been given concerning the replacement of HPLC by a certain assay, particularly for patients with hepatic dysfunction and those undergoing heart (HTx) or liver (LTx) transplantation (9)(10)(11)(12)(13)(14)(15)(16)(17)(18)(19)(20)(21)(22). Thus, it has remained difficult to achieve comparable results between transplantation centers.
In the present study, therefore, I evaluated the performance and particularly the specificity of all major monoclonal assays for CsA. Metabolite-to-parent ratios were estimated in all clinical specimens, using the TDx® polyclonal assay as a measure of the sum of metabolite and parent concentrations. The study includes RIA, TDx monoclonal (TDx-m), Emit with both pretreatments, AxSYM, and CEDIA from two clinical studies using HPLC as the reference method.
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
A total number of 145 randomly chosen samples from 78 patients after renal transplantation (KTx; 80 samples from 50 patients), LTx (63 samples from 26 patients), or combined transplantation (2 samples from 2 patients; LTx/KTx and KTx/HTx) were used in all assays. One hundred and eleven samples were obtained from outpatients, and 34 were obtained from inpatients.
The data of 6066 daily routine CsA measurements with Emit-MeOH and TDx polyclonal from 266 patients after LTx (1131 samples from 51 patients), KTx (4908 samples from 214 patients), and HTx (27 samples from 1 patient) were analyzed retrospectively. This provided data for the range of parent-to-metabolite ratios and thus the biases of monoclonal assays to be expected in large populations.
Data from a similar study (using RIA, Emit-MeOH, and TDx monoclonal and polyclonal assays) at a different transplantation center2 (transplantation center 2) allowed me to compare the results with those obtained for other types of transplantation and different CsA target values, and provided comparative data for the RIA. The previous 1992 study contained 613 samples from 166 patients after LTx (330 samples from 46 patients), KTx (94 samples from 46 patients), HTx (87 samples from 42 patients), bone marrow (80 samples from 27 patients), lung (6 samples from 1 patient), pancreas (3 samples from 1 patient), and combined transplantation (13 samples from 3 patients).
All patients had received dosages based on Emit-MeOH trough concentrations.
quality control
Precision studies were performed using controls supplied by
Bio-Rad Laboratories. External quality control was ensured through
participation in the Cyclosporin International Proficiency Testing
Scheme (Coordinator, Dr. D.W. Holt). Since April 1995, the maximum
deviation from the method means has been 1.8 SD, with an average of
0.42 SD for Emit-MeOH (mean bias, -0.06) and 0.58 SD for the TDx
polyclonal (mean bias, 0.4).
assays
The single-step whole blood CEDIA assay (Boehringer
Mannheim) was performed on a Hitachi 912 instrument in its final
marketed format. The monoclonal Emit-MeOH assay was supplied by Behring
Diagnostics Inc. and was performed on two Cobas Mira Plus instruments.
To evaluate the new pretreatment solution (Emit-NPT), we adhered to the
original manual pipetting procedure. The Abbott Laboratories AxSYM
monoclonal fluorescence polarization immunoassay and TDx-m use the same
proprietary antibody. The CycloTrac® specific
RIA from Incstar Corporation had been used in the previous 1992 study
and was performed strictly as recommended by the manufacturer.
The HPLC kit (ClinRep®) from Recipe Merck served
as a reference method specific for CsA, whereas the TDx polyclonal
(parent and metabolites; Abbott) gives an overall estimation of the
total concentration of CsA plus, to various degrees, several
metabolites of CsA (23) (Table 1
).
|
data analysis
Passing-Bablock analysis and Pearson correlation coefficients were
used to compare the results obtained with different assays. Only the
first result of duplicates was included. Wilcoxon–Mann–Whitney tests
were performed when different populations were compared for a
significant difference in means. Precision studies were done according
to NCCLS document EP5-T2. The differences in CVs
calculated from the duplicate patient results and reruns were tested by
the Wilcoxon matched-pairs signed-ranks test.
The overall content of metabolites in each sample was estimated by calculating "metabolic ratios" (MRs) between the results of the TDx polyclonal and HPLC (MR-H = TDx polyclonal/HPLC) or Emit (MR-E). The individual deviation from the HPLC (or Emit) result was then calculated for each monoclonal immunoassay in each sample and compared with the MR calculated in the same sample.
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Linearity.
There was a linear relationship between the
expected and the measured concentrations for all assays when the
highest calibrator was serially diluted: CEDIA; slope, 1.03 ±
0.02, intercept, -27.8 ± 6.9, Sy|x =
8.0;Emit-NPT: slope, 1.00 ± 0.02, intercept, -14.4 ± 5.0,
Sy|x = 6.2; TDx-m: slope, 0.93 ± 0.01,
intercept, -1.3 ± 2.1, Sy|x = 4.0;
AxSYM: slope, 1.03 ± 0.01, intercept, -4.3 ± 3.7,
Sy|x = 5.5.
Precision.
The total CVs of all control results during the
study period are shown in Fig. 1
A. The highest CVs were calculated for HPLC, the lowest over the
entire range for the TDx-m. Both Emit versions and the CEDIA gave
comparable CVs, although they were noticeably worse than the AxSYM at
concentrations <300 µg/L. The results of the duplicate
measurements of patient samples were grouped into concentration classes
of 20 µg/L. The average within-batch CV was then calculated for each
class (Fig. 1B
). In addition, all samples were tested in a rerun with
the Emit-NPT, the AxSYM, and the CEDIA, and total CVs were calculated
(Fig. 1C
). The lowest within-batch CVs were obtained with the TDx-m,
followed by both Emit versions and, lastly, the AxSYM (>100 µg/L)
and the CEDIA (<100 µg/L). There was no significant
difference between both Emit versions and the CEDIA. The AxSYM
performed significantly worse than both Emit versions
(P = 0.022 and 0.008) (8) and the CEDIA
(P = 0.024). Any other comparison between any two
assays yielded highly significant differences between their
within-batch CVs (P <0.0001). In contrast to the
within-batch CVs, the total CVs of the AxSYM were significantly lower
than both the Emit-NPT and the CEDIA when compared in patient samples
(P <0.0001).
|
Sensitivity.
The analytical limits of
detection, defined as 3 SD above the mean for the zero
calibrator (n = 10), were as follows: Emit, 20 µg/L;
AxSYM, 12 µg/L; and TDx-m, 8 µg/L (8). No zero
calibrator was available for the CEDIA. Instead, I used the lowest
calibrator (target value, 17.6 µg/L), which yielded a result of 31
µg/L. On the basis of a maximum tolerable CV of 10%
(4)(25), the functional sensitivity (limit of
quantification) was estimated as 60–70 µg/L for the Emit and CEDIA
and 40 µg/L for the AxSYM assay.
Recovery and cross-reactivity.
Four unknown samples (processed
base human serum with methylated human hemoglobin), provided by Abbott,
were tested with the Emit-NPT, AxSYM (20 replicates), CEDIA, TDx-m, TDx
polyclonal (10 replicates), and HPLC (4 replicates). CsA (target, 50 or
250 µg/L) had been added to each sample. The calculated recoveries
were as follows: Emit-NPT, 125.8% and 107.5%; TDx-m, 117.8% and
118.6%; AxSYM, 114.0% and 114.3%; CEDIA, 143.4% and 122.9%; TDx
polyclonal, 135.4% and 119.9%; and HPLC, 109.5% and 99.8%. The
recovery of the HPLC was 102.0% and 101.5% when CsA (125 or 270
µg/L) was added to CsA-free whole blood. Two samples also
contained 1000 µg/L metabolite AM1 and 500 µg/L AM9, respectively
(Novartis). The calculated cross-reactivities for AM1 and AM9 in the
presence of 250 µg/L CsA are shown in Table 1
(percentage of added metabolite appearing as additional
apparent CsA concentration).
conventional direct comparison of methods
The correlation coefficients and standard errors of the estimate
achieved (Table 2
) were better between the different monoclonal immunoassays
(r = 0.93–0.98) than between HPLC and immunoassays
(r = 0.89–0.93). This was attributable to the lower
CVs and the positive bias of the various immunoassays when compared
with HPLC (9–57%). The correlation coefficients with the polyclonal
TDx were much lower for all the monoclonal assays and for HPLC
(r = 0.63–0.78, data not shown).
|
Both Emit versions produced virtually identical results (y/x = 1.03). A good correlation had also been achievedwhen the Emit-MeOH assay was compared with the RIA for the 1992 data set (mean y/x = 0.99).
There was an unexpected difference when the results of the present study were compared with those from the 1992 study. A strong positive bias of the TDx-m when compared with the Emit-MeOH was found in the 1997 population (mean y/x = 1.44) as opposed to a small bias in 1992 (mean y/x = 1.12). It is essential to appreciate that there has been a shift in the relative performance of the Emit and TDx monoclonal over the last 5 years for unidentified reasons, possibly because of a shift in standardization of either of the two assays involved.
metabolite cross-reactivity and consecutive bias of monoclonal
assays
The comparison of MRs (as an estimate of metabolite-to-parent
ratios) with the observed positive biases of the different monoclonal
assays revealed a relatively strong correlation. Fig. 2
B suggests that the extent of the positive bias of the TDx-m
compared with the HPLC [bias = (TDx-m - HPLC)/HPLC] is
related to the amount of metabolites detected by the polyclonal assay.
At an MR-H of 10 (TDx polyclonal/HPLC), the calculated function
suggests a positive bias of 133%. On the other hand, samples with an
MR-H of 2 display a positive bias of only 20%. Fig. 2C
shows the
smaller positive bias for the AxSYM (96% at an MR-H of 10). The Emit
displays the lowest slope, followed by the CEDIA (Fig. 2
, A and D).
|
The Emit can, therefore, serve as an alternative reference method for
the calculation of MRs (MR-E = TDx polyclonal/Emit). Both MRs
correlated well [r = 0.832; MR-H = (1.16 ±
0.05)MR-E - (0.35 ± 0.20)]. After two outliers with HPLC
results close to the detection limit (21 and 25 µg/L) were
eliminated, the correlation coefficient was 0.885. The resulting slopes
and intercepts when the positive bias of the other monoclonal
immunoassays were correlated to MR-E are shown in Fig. 3
. These ratios (MR-E) were available in a much larger set of
samples and allowed the results of TDx-m and Emit from the present
study to be compared with those from 1992. The calculated correlation
coefficients and slopes and, thus, the dependence of TDx-m results on
metabolites (MR-E) were independent of the type of transplantation and
very similar for both transplantation centers (range of slopes,
7.4–9.8 for the various transplantations in both centers). The
differences of the intercepts between both transplantation centers,
producing an additional positive bias of 25%, resemble those of the
method correlation discussed above. They are consistent for all
transplantation types and support the idea that a change of
standardization rather than a change of metabolite cross-reactivity
could be responsible for the unexplained change of relative performance
between Emit-MeOH and TDx-m since 1992.
|
No significant correlation was found between MR-E and the RIA deviation from Emit-MeOH results (r = 0.175). This indicates no additional cross-reactivity for the RIA compared with the Emit.
type of transplantation and hepatic dysfunction
Can MRs and, consequently, the bias of monoclonal assays in
individual samples be predicted from the type of transplantation or
other identifiable factors? The mean MR-E observed in 1992 at
transplantation center 2 was 3.27 (n = 613) compared with 4.03 in
the present study population (n = 145) and 3.83 in the routine
samples (n = 6066). The observed mean MR-Es in the present study
as well as in the routine population were very similar for both KTx and
LTx (3.85 and 3.83) and for in- and outpatients (3.90 vs 3.76). The
observed average MR-Es at transplantation center 2 were as follows:
4.25 for HTx (range observed in different samples, 1.36–13.8); 2.32
for bone marrow transplantation (0.97–5.73); 3.19 for
LTx (0.95–13.5); and 3.54 for KTx (1.54–10.4). The
differences between the transplantations, although significant
(P <0.0001; exceptions were P =
0.02 for KTx vs HTx and P >0.05 for KTx vs LTx), were small
compared with the variability of individual results between each
subject or specimen irrespective of transplantation. The cumulative
frequency distribution of the metabolite-to-parent ratio estimates
(MR-E) of the three populations studied is shown in Fig. 3
, together
with the resulting positive bias of the TDx-m, AxSYM, and CEDIA as
compared with the Emit. There is a slight parallel shift towards lower
MR-Es in transplantation center 2, where significantly higher CsA
concentrations had been observed. In both transplantation centers,
higher relative metabolite concentrations were observed at lower parent
concentrations (Fig. 4
). Very high MRs (e.g., MR-E >12) usually occur at CsA
concentrations <100 µg/L. This is probably because of increased CVs
for the specific reference methods (HPLC or Emit) at low CsA
concentrations (26).
|
MR-Es are stable in individual patients over long periods of time. At
an average MR-E of 3.6 in 115 patients with at least five MR-Es
measured (mean, 30 MR-Es), the mean SD within a patient was only 0.7,
including the immediate post-transplantation phase. The whole
population of patients, however, covered a wide range of individually
kept MR-Es (2.1–7.2), irrespective of transplantation (Fig. 5
), which pointed to the importance of genetic predisposition
rather than environmental influences. Consequently, standard
biochemical liver tests do not safely indicate MRs. Table 3
shows the three patients with the highest and the one
patient with the lowest MR-E, the highest 
-glutamyl transferase (EC
2.3.2.2), the two highest alkaline phosphatase (EC 3.1.3.1), and the
highest bilirubin values. When the correlation between 106 bilirubin
and 107 -glutamyl transferase results available in the current study
population and the MR-Es was calculated, it yielded very weak
correlation coefficients of 0.315 and 0.390.
|
|
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
No significant difference was observed when the patient results of both Emit versions were compared. Within-batch precision was second only to the TDx-m. The total CVs were significantly worse than those for the AxSYM and TDx-m methods, probably because of insufficient calibration curve stability. It has been suggested that CV improvement should be expected on automated systems because there would be fewer steps involving liquid handling (7). This would primarily improve within-batch precision. No influence can be expected on the high total CVs caused by calibration instability.
The new AxSYM CsA assay performed very well, but showed a positive bias
of 29% compared with HPLC and 16% compared with the Emit-NPT because
of cross-reactivity. This is a definite improvement in specificity over
the TDx-m (57%) and has been achieved by the modification of some of
the assay conditions. However, interpreting results from different
laboratories becomes even more difficult with yet another method not
specific for the CsA parent on the market (8). The
performance of the assays is summarized in Table 4
.
|
The specificity of monoclonal immunoassays for CsA has been addressed in the past mostly by direct comparison with HPLC or by measurement of purified metabolites. The former provides data about the mean bias of the population examined; the latter provides necessary basic information. Both methods are of limited value for explaining the different biases of monoclonal assays observed in single samples. Few data correlating the presence of metabolites in clinical samples and the irregular biases observed with monoclonal immunoassays are available. This is because HPLC procedures for measuring metabolite concentrations are labor-intensive and time-consuming. Ratios between nonspecific immunoassays and specific methods have been widely used to estimate overall metabolite concentrations in patient samples and to identify differences between certain populations. Large interindividual differences have also been observed, and the use of nonspecific assays has been discouraged because of these results (27)(28)(29)(30)(31). Nonspecific immunoassays in combination with specific HPLC have seldom been used to assess the different biases observed in single clinical specimens when CsA is measured with monoclonal assays. These ratios are certainly inferior to HPLC for studies on CsA metabolism. Nevertheless, it is obvious from experimental studies (17)(20)(24) and theoretical consideration that the cross-reactivity of monoclonal antibodies is primarily detectable against the structurally related first-generation metabolites that are altered in only one position. Usually, AM1 and AM9 are also the most abundant metabolites in whole blood, with the AM1 concentration equaling or even exceeding that of CsA (2)(23)(32)(33). The polyclonal TDx assay primarily detects first-generation metabolites (2) and is, therefore, useful for studying the metabolite cross-reactivity of monoclonal assays.
This study demonstrates that MRs (TDx polyclonal/HPLC or TDx polyclonal/Emit) correlate with the positive bias observed in all investigated monoclonal immunoassays as compared with the more specific method, including the latest developments with reduced cross-reactivity (TDx-m, AxSYM, CEDIA, and Emit). The individual bias observed in a sample depends predominantly on the individual content of metabolites. The data show that the cross-reactivity of all monoclonal assays, despite appearing negligible at first glance, can lead to substantial overestimation of the true CsA concentration in a number of patients. These patients could not be easily identified as being only patients with liver dysfunction or undergoing certain transplantations. The correlations and slopes describing the overestimation in single specimens were very similar when compared between two transplantation centers with different CsA target concentrations. A similar function has been reported for the TDx-m and TDx polyclonal/RIA (specific) ratios (12).
The TDx polyclonal detects AM1 markedly better than AM9, thus explaining the higher correlation coefficients for the deviation of both TDx-m and AxSYM compared with those of Emit-MeOH/NPT and CEDIA. According to the results from purified metabolites, AM1 is the major problem for both monoclonal Abbott assays, whereas the other two assays cross-react more strongly with AM9 [manufacturers' inserts and Refs.(17)(20)].
Like many other authors (12)(27)(30)(34)(35), I
found highly significant differences in the mean MRs observed for
different transplantations. This indicates a higher risk of metabolite
accumulation, particularly in HTx. Nevertheless, the maximum difference
observed for the various transplantation types (MR-E: bone marrow
transplantation, 2.32; HTx, 4.25) is moderate compared with the large
differences observed in individual samples from any transplantation
(27)(28)(29)(30)(31)(36)(37)(38)(39). Consequently, the same applies to the
different biases observed in monoclonal assays. Not only single
specimens but also the long-term mean MRs, which reflect an
individual's genetic predisposition, showed more variability than was
introduced by the type of transplantation (Fig. 5
) (29). The
large interindividual differences agree with studies showing that the
catalytic activities of cytochrome P450 3A, the enzyme family
responsible for the formation of first-generation metabolites
(2), vary by at least 10-fold (40)(41).
Similar to earlier reports (37)(42)(43)(44), a very weak
correlation between bilirubin and -glutamyl transferase and
increased MRs was detectable in our population. Temporary increases in
metabolite concentrations have been described, in particular, with
severe hepatic dysfunction and immediately after transplantation
(10)(27)(35)(36)(45). Lacerda et al. (46),
however, found no correlation between hepatic metabolite concentrations
and either serum bilirubin or the degree of cholestasis in liver biopsy
specimens. Bleck et al. (33) and Christians et al.
(39) showed that cholestasis was associated with a selective
increase in concentrations of second-generation metabolites (AM19,
AM1c9, AM1A, and AM11d). No differences in the concentrations of
first-generation metabolites were observed in that study. Consequently,
monoclonal assay results were not influenced (3H-RIA) as in another
study assessing Emit and TDx-m (26). The conclusion by
Witzke et al. (26) that the contribution of metabolites to
the TDx-m signal is predictable and constant is not justified. It
ignores the presence of interindividual differences irrespective of
hepatic function. To summarize the literature, there is convincing
evidence for an association between hepatic dysfunction and increased
predominantly second-generation metabolites. These are only partly
detected by the TDx polyclonal and hardly pose a problem of
cross-reactivity for monoclonal antibodies. This weak association is
also superimposed by a large interindividual variability, which is
probably determined genetically. Consequently, standard biochemical
liver tests were unable to indicate the individual bias of monoclonal
assays in our population. High MRs could be observed with either normal
or abnormal hepatic function as reported by Tredger et al.
(43). Increased ratios may be anticipated, however, under
co-therapy with interfering substances (2)(47)(48).
The distribution of MRs shows that 5% of all patients had an MR-E >6.4 and thus a deviation from the Emit >64% with the TDx-m and 37% with the AxSYM assay. Samples from patients with low MR-Es showed little or no bias. The resulting bias is higher when HPLC is used as the specific reference method (TDx-m >92%, AxSYM >61%, CEDIA >36%, Emit >21%). From this study, it is obvious that the average bias obtained from conventional method correlation studies is only valid for patients with average MRs. The CsA concentration in all other patients will be over- or underestimated. Reference ranges have been established, mostly using specific HPLC methods (49). The suggestion to modify reference ranges according to the results of method correlation studies (10) does not seem to be justified. The practice of adopting new methodology and reporting reference ranges 0–67% higher than the one for HPLC (6) disregards the different bias shown for every patient. Measurements by one technique cannot be adjusted to allow use of a therapeutic range determined for another method (4)(27)(31)(37). According to the results of this study, this is also true for monoclonal CsA assays.
The individualization of drug dosages is usually done by adjusting a standard dose e.g., according to the body weight. If additional individualization is required, therapeutic drug monitoring can usually help to achieve safe, sufficient drug concentrations in each patient. This is only possible, however, if the measured value is close to the true value or shows the same bias for all patients. The differences in bias for different patients in this study exceeded all other factors of influence described in the literature and sometimes even exceeded the published therapeutic ranges (Emit vs HPLC, -7% to 53%; TDx-m vs HPLC, 20–174%; AxSYM vs HPLC, 2–130%; CEDIA vs HPLC, -3% to 81%). The problem with cross-reactivity is not the overall bias observed, but the extended range of biases. The following illustrates this point: a result of 150 µg/L obtained by TDx-m in an individual sample can mean a true value (HPLC) of 55 or 125 µg/L when the two extremes observed in our 145 sample population are used. This is comparable to giving the same dose of a drug that is usually individualized by body weight to both a 55-kg and a 125-kg patient. The analytical error must be added on top of that. Under such circumstances, one must question whether drug-monitoring of CsA is justified at all. It at least implies that inappropriate dosage adjustments could be made from the use of such methods (50).
Except for the immediate post-transplant period in HTx and LTx patients, it has been advocated that TDx-m results satisfactorily parallel those of HPLC (10)(20). Few data were available to confirm or deny this assertion (22). Adding to the confusion are the many different biases, slopes, and intercepts that have been published in method comparison studies, even those comparing the same methods, in particular for the TDx-m assay (9)(22)(51). Cross-reactivity to CsA metabolites has been identified as the principal cause of these biases, and doubt has been raised early about the equivalence of HPLC and monoclonal immunoassays (38). Separate biases for different transplantations have been calculated (10)(11)(12). Apart from misuse of the terms bias and slope (22)(52), this is probably because of the presence or absence of samples with high MRs and the concentrations at which these are observed. In the present study, the addition of just five theoretical results with an MR-E of 10 to the study population (assumed concentration either 50 or 200 µg/L) caused a change in slope and intercept of 0.1 and 10 (TDx-m vs Emit-NPT, n = 145). Both the number of samples with high metabolite-to-parent ratios and the CsA parent concentrations at which these are observed are highly influenced by chance in small study populations. In a random sample of 100 specimens, the number of specimens with an MR-E >6.4 varies between 2 and 11 (95% confidence interval; expected mean, 5). When the weight of these specimens on the slopes and intercepts of regression curves is considered, many of the varying results reported for comparisons of the same assays in different studies can be explained.
The extent of the contribution of CsA metabolites to immunosuppression or toxicity has been discussed at length (1)(2)(22)(53). Presently, monitoring of metabolites is not recommended as a standard procedure (6). Specific assays should be used if assessment of the immunosuppressive activity or toxic potential of the metabolites is deemed necessary (20). It could not be shown that specific assays could predict acute rejection better than polyclonal assays (54)(55)(56). However, the consensus is that analytic methods used should be specific for CsA because of the well-defined activity and toxicity of CsA and the improved interlaboratory comparison, which results from measuring standardized, chemically defined structures. This point of view is strongly supported by the results of this study.
The results of this study also imply a possible influence of assay technology on pharmacokinetic studies (22) and in randomized concentration-control clinical trials (57) and are, therefore, at variance with Aspeslet et al. (58), who stated that specific fluorescence polarization immunoassay and RIA methods provided valid results irrespective of their cross-reactivity.
To conclude, there is still a need for an easy, fast, and truly
specific assay with high precision and sensitivity. The consensus
guidelines for assessing the specificity of CsA assays
(4)(6) (slope to HPLC 
10% from the line of identity;
intercept 15 µg/L, Sy|x 15 µg/L)
should be revised because they do not ensure sufficient specificity. It
is necessary to demonstrate that there is no correlation between any
observed bias and the individual content of metabolites in patients.
Similar problems may be encountered in the monitoring of other drugs
[e.g., Tacrolimus (59)(60)].
|
Acknowledgments |
|---|
|
Footnotes |
|---|
1 Nonstandard abbreviations: CsA, cyclosporin A; Emit-NPT, Emit with new pretreatment; HTx, heart transplantation; TDx-m, TDx monoclonal; LTx, liver transplantation; Emit-MeOH, Emit with methanol extraction; KTx, kidney transplantation; MR, metabolic ratio; MR-H, metabolic ratio with HPLC as reference method; and MR-E, metabolic ratio with Emit as reference method. 
2 2 Analyses were done by myself and the same technicians as in the present study at the Institute of Clinical Chemistry, Klinikum Grosshadern, Ludwig-Maximilian University, Munich, Germany.
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The following articles in journals at HighWire Press have cited this article:
|
|
 |
|
  N. Karamperis, P. Koefoed-Nielsen, A. Bagger Sorensen, C. Hojskov, J. Poulsen, and K. Jorgensen Role of metabolites and calcineurin inhibition on C2 monitoring in renal transplant patients Nephrol. Dial. Transplant., March 1, 2005; 20(3): 618 - 621. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 R. G Morris Immunosuppressant Drug Monitoring: Is the Laboratory Meeting Clinical Expectations? Ann. Pharmacother., January 1, 2005; 39(1): 119 - 127. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. R. Terrell, T. M. Daly, K. G. Hock, D. C. Kilgore, T. Q. Wei, S. Hernandez, D. Weibe, L. Fields, L. M. Shaw, and M. G. Scott Evaluation of a No-Pretreatment Cyclosporin A Assay on the Dade Behring Dimension RxL Clinical Chemistry Analyzer Clin. Chem., July 1, 2002; 48(7): 1059 - 1065. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
F. Streit, V. W. Armstrong, and M. Oellerich Rapid Liquid Chromatography-Tandem Mass Spectrometry Routine Method for Simultaneous Determination of Sirolimus, Everolimus, Tacrolimus, and Cyclosporin A in Whole Blood Clin. Chem., June 1, 2002; 48(6): 955 - 958. [Full Text] [PDF]
|
|
|
|
|
|
B. G. Keevil, D. P. Tierney, D. P. Cooper, and M. R. Morris Rapid Liquid Chromatography-Tandem Mass Spectrometry Method for Routine Analysis of Cyclosporin A Over an Extended Concentration Range Clin. Chem., January 1, 2002; 48(1): 69 - 76. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
D. W. Holt, A. Johnston, B. D. Kahan, R. G. Morris, M. Oellerich, and L. M. Shaw New Approaches to Cyclosporine Monitoring Raise Further Concerns about Analytical Techniques Clin. Chem., June 1, 2000; 46(6): 872 - 874. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||
), HPLC; (
), RIA; (
),
Emit-MeOH; (
), Emit-NPT; (
), TDx-m; (
), AxSYM;
(
), CEDIA.
