Effects of Anticoagulants in Amino Acid Analysis: Comparisons of Heparin, EDTA, and Sodium Citrate in Vacutainer Tubes for Plasma Preparation

Measurements of the concentrations of free amino acids in physiological fluids such as blood, cerebrospinal fluid (CSF), and urine are used mainly as biochemical indicators of inborn errors of metabolism (aminoacidopathies), nutritional status, and as monitors of therapy (1). Amino acids can be measured by thin-layer chromatography, electrophoresis, mass spectrometry, and ion-exchange chromatography (2). The last method has many advantages: it provides good resolution, is very accurate, can be automated, and is easy to perform. The long analytic time is the only drawback in this method. The steps in plasma free amino acid analysis, including sample pretreatment, deproteinization, and instrument operation, are easily performed. However, the use of different anticoagulants for plasma sample preparation can increase the potential for analytic errors (2)(3). Currently, heparin and EDTA are widely used as anticoagulants, but sodium citrate-treated plasma and serum samples are occasionally used. We investigated the analytic variation among samples treated with these anticoagulants. We used a System 6300 high performance amino acid analyzer (Beckman Instruments), which uses a single-column, ion-exchange method that uses four lithium buffers (Li-ADEF method), a ninhydrin reagent, and a 12-cm lithium high performance column. Four different types of Vacutainer Tubes were purchased from Becton Dickinson, including a plasma tube with heparin (sodium heparin), a whole blood tube with liquid EDTA (K3, 15%), a whole blood tube with 0.129 mol/L buffered sodium citrate (3.8% sodium citrate), and a serum separation tube. Centrifuge filtration was performed to eliminate the residual protein in sulfosalicylic acid dihydrate (SSA) supernatants. The microspin filter units (0.2 μm) were from Lida Manufacturing. All lithium buffers, ninhydrin reagents, and calibrators were purchased commercially from Beckman Instruments. Four lithium buffers with various pHs were used in this experiment: Li-A buffer (pH 3.0), Li-D buffer (pH 3.0), Li-E buffer (pH 3.7), and Li-F buffer (pH 4.0). The ninhydrin reagent was prepared by mixing 18.0 g of ninhydrin and 0.7 g of hydrindantin with a reagent containing (per liter) 767 mL of dimethyl sulfoxide, 225 mL of water, 7.0 g of lithium acetate, and 1 mL of concentrated acetic acid; the reagent was kept at room temperature for 4 h. Calibrators and quality control materials were made by mixing 400 μL of STD Calibration Standard, 400 μL of AN Calibration Mixture, and 400 μL of B Calibration Mixture (Beckman Instruments) with 400 μL of 25.0 mmol/L glutamine–asparagine, 400 μL of 5.0 mmol/L tryptophan (Trp), and 400 μL of 5.0 mmol/L d-glucosaminic acid (an internal standard) (all from Sigma Chemical) in a 10-mL volumetric flask and bringing the volume to 10 mL with Li-S sample dilution buffer. To deproteinate the plasma, we used an aqueous solution of 35 g of SSA (Sigma Chemical) in a total volume of 100 mL. To help separate and identify the amino acids, the plasma samples treated with SSA had the same pH as the amino acid calibration mixture used to calibrate the amino acid analyzer (2). Blood samples were taken from 43 healthy adults (15 men and 28 women; 23–48 years of age) after a minimum of 12 h fasting. All subjects gave signed informed consent. The results of all biochemistry tests, such as aspartate aminotransferase, alanine aminotransferase, total protein, and albumin, were within the reference intervals in these 43 adults. No patients with aminoacidopathies were included in this study. Blood samples were drawn in 4.5- or 7-mL heparin, EDTA, sodium citrate, and serum separation Vacutainer Tubes. The plasma was separated from the blood cells immediately after the blood was drawn by centrifugation at 2000g for 15 min. A plasma sample of 200 μL was placed in a 0.6-mL centrifuge tube, and 20 μL of SSA solution (350 g/L) was then added. The tube was capped, agitated for a few seconds, allowed to stand for 5 min at room temperature, and then centrifuged at 11 200g for 5 min. The SSA supernatant was transferred to a microspin filter reservoir (0.2 μm), capped, and centrifuged at 1800g for 30 min. The centrifuged aliquot was diluted with an equivalent volume of Li-S sample dilution buffer containing 0.256 mol/L of d-glucosaminic acid, which was used for quality control. Finally, the prepared sample (50 μL) was automatically injected into the analyzer for free amino acid analysis. We used the Li-ADEF method (Beckman) with care to keep the flow rate (20 mL/h), temperatures (33.0 °C, 65.5 °C, and 73.0 °C), and times of buffer transitions constant to avoid fluctuations in the retention times of the amino acids. The mean amino acid concentrations (± SD) for adults varied among sample types (Table 1). The concentrations of most free amino acids were higher in serum than in plasma. The sodium citrate-treated samples yielded the lowest values of most free amino acids. Results from both of these sample groups differed significantly from the heparin-treated control group, based on paired t-tests (t-values: serum, 2.88; sodium citrate, 2.38; P <0.005). No statistically significant differences were seen between EDTA-treated samples and the heparin-treated controls. Normal values for plasma free amino acids (μmol/L) for four anticoagulants. n = 43 healthy adults, 15 men and 28 women; ages 23–48 y. Range, mean ± SD. Normal values for plasma free amino acids (μmol/L) for four anticoagulants. n = 43 healthy adults, 15 men and 28 women; ages 23–48 y. Range, mean ± SD. The CVs for most free amino acids of the four sample types (n = 172) were between 12% and 42%, except for the following amino acids: phosphoethanolamine (PTEN), aspartic acid (Asp), hydroxyproline (Hyp), citrulline (Cit), β-alanine, ethanolamine, and Trp. The high CV found in the above amino acid quantitations came mainly from anticoagulant effects, which might have produced relatively poor resolution and incongruous retention times. Serum samples gave an average increase of 27.8% (range, 10–54%) in most free amino acids, except for taurine, Asp, Hyp, and Cit, for which the degrees of variation were all >100%. Remarkably, we found the same cystine (Cys) concentration in serum as in the heparin-treated control. This was not true of either the EDTA- or sodium citrate-treated samples. Sodium citrate-treated plasma had an average decrease of 21.3% (range, 8–40%) in most free amino acids, except for the dicarboxylic amino acids, Asp and glutamic acid, which were increased 38% and 23%, respectively. The concentration of Cys was about 10% of that of the control. EDTA-treated samples showed few variations from the control, except for PTEN, Cit, Cys, and Trp. The PTEN concentration was 20 times higher than in any of the other samples. The concentration of Cit was also markedly increased in EDTA-treated samples, and the variation was too large to be accurately determined. A very low concentration of Cys was also observed in EDTA, about 3% of the concentration of the control group. This was very close to the concentration measured in the sodium citrate-treated samples. In the free amino acid analysis, the most variable retention time was the interval between Asp and serine, ranging from ∼11.10 (± 2.5) min to 15.10 (± 1.0) min. When serum or sodium citrate-treated plasma was compared with control heparin-treated plasma, a delayed retention time of, on average, 0.56–0.78 min was observed between Hyp and threonine, respectively. Nevertheless, there was no substantial difference when EDTA-treated plasma was used (Fig. 1). The average within-run imprecision (CV) under these four conditions was 2.0% (range, 1.3–2.6%) for all free amino acid (n = 16) multiple measurements. Retention time evaluations comparing serum, sodium citrate-, and EDTA-treated plasma with heparin-treated plasma as a control. (A) A delayed retention time was found between threonine and serine in the serum and sodium citrate-treated groups. (B) No substantial differences were found between EDTA and heparin-control groups. Although serum is the most commonly used sample in many biochemical tests, plasma is preferred over serum for amino acid analysis. When blood clots, amino acids may be released from or metabolized by erythrocytes, leukocytes, or platelets and are then dispersed into the serum. Thus, serum has relatively higher and more variable concentrations of most free amino acids than does plasma (1)(2)(4)(5). According to our investigation, taurine, Hyp, Asp, Cit, and arginine showed a wide fluctuation of concentration in serum. In addition, the concentrations of glutamine and asparagine were slightly increased. This latter finding did not match the results of Scriver et al. (1), which showed decreased concentrations of glutamine and asparagine. The results from EDTA-treated samples were fairly comparable with the results from heparin-treated samples. However, we found that caution must be exercised in evaluating results from these samples. EDTA can react with the ninhydrin reagent and produce a ninhydrin-positive contaminant that is eluted from the column at 1.60–2.52 min. This contaminant might be added to the concentration of any amino acid with the same retention time by mistake, giving a falsely increased value. We found a large increase in the quantity of PTEN in EDTA-treated plasma, about a 20-fold increase over the other sample groups. The retention time appeared to have a 0.4-min delay on average. This possible source of error should be kept in mind if EDTA-treated samples are used. Heparin contains sodium metabisulfite as a preservative. This interferes with the quantitation of sulfur-containing amino acids, such as cystine and homocystine, by adding sulfocysteine to the sample (6). In our study, the cystine measured in heparin-treated samples was much higher than in sodium citrate- and EDTA-treated samples. No substantial difference in cystine concentrations was observed between serum and heparin-treated plasma. Thus, sodium citrate- and EDTA-treated samples may yield more accurate values for cystine. In summary, the anticoagulant used in sample preparation may markedly affect the results of amino acid analysis by automated ion exchange chromatography. Sample preparation should be standardized, and clinicians should be aware of the possible sources of error. We recommend the use of either heparin or EDTA exclusively, with careful attention to potential inaccuracies in cystine concentrations with the former and PTEN with the latter. fax 886-2-25433642, e-mail zsplin@ms2.mmh.org.tw We greatly appreciate the Department of Medical Research, Mackay Memorial Hospital, which provided the necessary instrumentation and financial assistance to complete this project. This project was partly supported by Research Grant No. 8519, Mackay Memorial Hospital, Taipei, Taiwan.