| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Technical Briefs |
1 Department of Clinical Chemistry, BovenIJ Hospital, Statenjachtstraat 1, 1034 CS Amsterdam, The Netherlands2 Department of Clinical Chemistry, Medical Center Alkmaar, Alkmaar, The Netherlands3 Department of Clinical Chemistry, Elkerliek Hospital, Helmond, The Netherlands4 Department of Clinical Chemistry, Gemini Hospital, Den Helder, The Netherlands5 Beckman Coulter Netherlands B.V., Mijdrecht, The Netherlands
aauthor for correspondence: fax 31206346529, e-mail h.huijgen{at}bovenij.nl
Homocysteine (Hcy) is a sulfur amino acid and a metabolite of the amino acid methionine. Hcy is very reactive because it contains a free sulfhydryl (thiol) group, and it readily oxidizes to form various disulfides. The fraction of free Hcy in plasma is therefore <2% of total plasma Hcy (tHcy) (1).
Increased Hcy has been associated with cardiovascular, cerebrovascular, and peripheral vascular disease (2)(3)(4) and is recognized as an independent risk factor. The need for a simple automated assay is therefore increasing, and various analytical methods have become available. Currently the two most widely used techniques are HPLC and immunochemistry.
We implemented a new automated enzymatic method for the measurement of tHcy on two different routine clinical chemistry analyzers and compared these assays with HPLC and the AxSYM immunoassay.
Reagents and calibrators for the enzymatic method were obtained from Catch Inc. In this method Hcy and L-serine form cystathionine, which is then converted to Hcy, pyruvate, and ammonia. The enzymes involved are cystathionine ß-synthase and cystathionine ß-lyase, respectively. tHcy is measured by the production rate of pyruvate by inclusion of lactate dehydrogenase and NADH in the reaction mixture. Reagent 1 contains serine, NADH, and lactate dehydrogenase; reagent 2 contains the reducing substance; and reagent 3 contains cystathionine ß-synthase and cystathionine ß-lyase. The calibrators have Hcy concentrations of 0.0 and 26.5 µmol/L (5). The enzymatic assay was implemented on two Synchron LX-20 and CX-5 analyzers (Beckman Coulter, Inc.).
The HPLC method is a modification from Spaapen et al. (6), as obtained from Instruchemie. Plasma samples were pretreated with tri-n-butylphosphine for reduction and deproteinized with trichloroacetic acid. Derivatization was performed with 7-fluorobenzo-2-oxal,3-diazole-4-sulfonic acid. 2-Mercaptoethylamine was used as an internal standard. Separation was obtained by use of an Inertisil ODS-3 column with sodium acetate buffer as the mobile phase. Assay recovery was optimized by use of samples obtained from the Dutch Foundation for Quality Assessment in Medical Laboratories.
The immunoassay is based on the fluorescence polarization immunoassay technology of the AxSYM immunochemistry analyzer (Abbott Laboratories) (7).
A total of four Dutch clinical chemistry laboratories participated in the study. Laboratory 1 was the laboratory of the BovenIJ Hospital (315 beds) in Amsterdam, equipped with both the CX-5 and LX-20 analyzers. Laboratory 2 was the laboratory of the Medical Center Alkmaar (800 beds) in Alkmaar, equipped with the CX-5 analyzer and a HPLC. Laboratory 3 was the laboratory of clinical chemistry of the Elkerliek Hospital (524 beds) in Helmond, equipped with the LX-20 analyzer. Laboratory 4 was the laboratory of clinical chemistry of the Gemini Hospital (320 beds) in Den Helder, equipped with the AxSYM. All laboratories participated in the comparison study, and laboratories 1, 2, and 3 participated in the analytical evaluation.
Samples for testing linearity were prepared in laboratories 1, 2, and 3, and samples for the interference study were prepared in laboratories 1 and 3. Samples for the comparison study were collected in laboratory 2 from patients screened for tHcy by a fasting plasma measurement and/or an oral methionine loading test. Samples (anonymized) were used for the study only if patients gave their informed consent, according to the guidelines of the board of directors of the hospital. A total of 100 samples were used. The patient demographics were as follows: males, n = 42 patients; mean age, 50 years (range, 32–70 years); mean tHcy, 19.5 µmol/L (range, 5.3–110.6 µmol/L); females, n = 58 patients; mean age, 44 years (range, 20–79 years); mean tHcy, 20.2 µmol/L (range, 4.1–100.7 µmol/L). Blood was drawn in tripotassium EDTA tubes (Vacutainer; Becton Dickinson BV) and kept on ice; plasma was separated within 30 min and stored at –20 °C. The frozen aliquots were distributed, thawed, and measured in the participating laboratories on the same days.
Imprecision was assessed in laboratories 1, 2, and 3 by use of commercial control samples, all prepared from human serum, obtained from three suppliers: Bio-Rad Laboratories BV; UTAK Laboratories Inc.; and the Dutch Foundation for Quality Assessment in Medical Laboratories.
Within-run imprecision (CV) was determined by measuring six different control samples 21 times in one run. Total imprecision was established according to the EP5 protocol (8) and linearity according to the EP6 protocol (9). The lower limit of detection was defined as the concentration corresponding to a signal 3 SD above the mean (n = 20) for calibrator 0, which was free of Hcy. The influence of bilirubin, hemoglobin, and lipids was evaluated according to the CERMAB protocol (10).
We compared the results obtained with the three methods by debiased regression according to Passing and Bablok (11). The mean difference was calculated by the paired t-test and displayed graphically as recommended by Bland and Altman (12). All statistical analyses were done with Analyze-It software.
The enzymatic assay was linear from 1.5 to 90.0 µmol/L on both the LX-20 and CX-5 analyzers. The lack of linear fit test was rejected at P <0.01. The limits of detection on the LX-20 and CX-5 were 0.21 and 0.41 µmol/L, respectively. The results from the imprecision study are presented in Table 1
.
|
Interference studies showed that hemoglobin up to 0.97 mmol/L (1600 mg/dL) and bilirubin up to 267 µmol/L (15.6 mg/dL) did not interfere in the assay. Triglycerides >9.7 mmol/L produced a nonlinear significant decrease in measured tHcy concentrations of more than twice the total imprecision of the test on both analyzers.
Bland–Altman plots comparing the enzymatic methods on both the CX-5 and the LX-20 with HPLC are shown in Fig. 1
. Passing–Bablok regression analysis revealed that both the LX-20 and CX-5 measured tHcy concentrations that were significantly higher than HPLC results. The calculated slopes (95% confidence intervals) for the LX-20 in laboratories 1 and 3 were 1.060 (1.04–1.085) and 1.034 (1.014–1.053), respectively, and for the CX-5 in laboratories 1 and 2 were 1.047 (1.030–1.065) and 1.022 (1.004–1.042), respectively. For all comparisons, the intercepts were not significantly different from 0.0 µmol/L. When compared with the AxSYM, the calculated slopes for the LX-20 in laboratories 1 and 3 were 1.078 (1.054–1.100) and 1.044 (1.014–1.073), respectively, and for the CX-5 in laboratories 1 and 2 were 1.076 (1.049–1.097) and 1.050 (1.031–1.076), respectively. The calculated intercepts were –0.322 (–0.575 to –0.009), –0.289 (–0.715 to 0.010), –0.461 (–0.702 to –0.163), and –0.485 (–0.793 to –0.252) µmol/L, respectively.
|
Until recently the measurement of tHcy in plasma was limited to laboratories equipped with highly specialized instrumentation such as gas chromatography–mass spectrometry or HPLC. The introduction of an immunoassay (7) brought homocysteine testing within reach of routine clinical chemistry laboratories. Recently a new enzymatic tHcy test has become commercially available that is easy to implement on a routine clinical chemistry analyzer, which is preferable from a logistic point of view.
The linear range of the enzymatic assay is adequate. Samples with tHcy concentrations >90 µmol/L should be diluted, but concentrations that high are seldom found in clinical practice.
The within-run imprecision was good with exception of the LX-20 measurements at low concentrations in laboratory 1. The results obtained with the same tHcy concentration in laboratory 3 led us to conclude that the performance was analyzer dependent and thus can be optimized.
Total imprecision (CV) ranged from 2.5–3.5% (high controls) to 4.1–7.0% (low controls). According to Fraser et al. (13), objective goals for total imprecision can be calculated based on the CVwithin-subject. They proposed a desirable performance for imprecision as 0.5 x CVwithin-subject. The reported within-subject variation of tHcy is 7.0–40%, with a value of 
8.5% reported most frequently (14)(15)(16)(17)(18)(19). It therefore can be concluded that the total imprecision of this enzymatic assay does not meet the criterion of desirable performance at all concentrations. Several studies on interlaboratory variation of tHcy measurements have been published (20)(21)(22)(23). Compared with the results reported in those studies, the total imprecision of the enzymatic method seems to be better than that for HPLC but somewhat worse than the imprecision of the AxSYM assay.
From the patient-comparison study it can be concluded that both the CX-5 and LX-20 analyzers report higher tHcy concentrations than HPLC, whereas the AxSYM results are lower. None of the calculated intercepts for regression comparison of the enzymatic assays with HPLC were significantly different from 0.0 µmol/L. This highlights the fact that the differences between the two methods are probably caused by differences in the calibrators, which is supported by the observation that with the HPLC method, the measured tHcy concentration of standard solution 2 of the enzymatic test was 7.5% higher than the value stated by the manufacturer. Standard solution 1 did not contain Hcy.
In Fig. 1
two data points catch the eye, at 43.9 and 80.8 µmol/L tHcy. For both data points, the tHcy values measured by the LX-20 and CX-5 analyzers were much higher than the results obtained by HPLC and the AxSYM. These two samples were obtained after methionine loading, and the tHcy concentrations of the matching fasting samples were very similar with all three formats. After a methionine load the Hcy metabolism is stressed in a nonphysiologic way, potentially leading to increased concentrations of intermediates of the methionine Hcy pathway. In our study 26 paired samples from a methionine load test were collected. To determine whether methionine loading influences the enzymatic measurement, we compared the differences (expressed as the percentage relative difference) between the enzymatic assay and HPLC for the fasting samples with the differences between the enzymatic assay and HPLC for the post-methionine load samples. We found no significant difference between the fasting differences and post-methionine load differences [t-test, fasting mean (SD), 6.12 (6.59)%; post-methionine load mean (SD), 6.49 (6.64)%; difference between the means, –0.37% (P = 0.805; n = 26)].
In conclusion, from the point of practicability, this new, easy to implement enzymatic method is useful in a routine clinical chemistry laboratory setting. However, to meet the criterion of desirable performance, total imprecision could be improved by measuring the plasma samples in replicate. Nevertheless, compared with the imprecision of the widely used HPLC method, the enzymatic method is more precise. Moreover, the tHcy concentrations measured by this assay are slightly but significantly different from those measured by HPLC and immunochemistry, although we found a strong linear correlation between the methods. With regard to the small differences found, calibration of the test with samples provided by Institutes for External Quality Assessment could easily correct this problem.
Acknowledgments
We thank Beckman Coulter Netherlands B.V. (Mijdrecht, The Netherlands) and Catch Inc. (Seattle, WA) for providing us with reagents for the enzymatic assay and Dr. A.P. van Zanten, clinical chemist, for comments.
References
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
