Serum protein electrophoresis by CZE 2000 clinical capillary electrophoresis system

^a Address correspondence to this author at: Central Clinical Laboratory, Clinical Pathology Department, University Hospital Leuven, Kapucijnenvoer 33, B-3000 Leuven, Belgium. Fax 32 16 332896; e-mail xavier.bossuyt{at}uz.kuleuven.ac.be.

	Abstract

Top Abstract Introduction Materials and Methods Results and Discussion References

 
We compared the automated Paragon 2000 clinical capillary zone electrophoresis (CZE) system with two manual methods, agarose electrophoresis (AGE) and cellulose acetate electrophoresis (CAE). Reference intervals in healthy adults were determined for each method. When compared with AGE and CAE, CZE gave substantially higher reference values for the ₁-globulin fraction. With CZE, within-run precision for fraction quantitation was between 0.5% (albumin) and 4.1% (₁-globulin). Total precision was between 0.8% (albumin) and 5.3% (ß-globulin). Data obtained from CZE showed poor linear correlation with results obtained by AGE but good linear correlation with data from CAE. Analysis of serum from patients with inter alia inflammation, nephrotic syndrome, or polyclonal gammopathy showed that clinical information obtained by CZE is comparable with information obtained by AGE and CAE. We conclude that CZE offers a clinically reliable alternative to AGE and CAE and has the advantages of automation, higher precision, and faster turnaround time.

	Introduction

Top Abstract Introduction Materials and Methods Results and Discussion References

 
Electrophoresis of serum proteins has been an established and effective disease-screening tool in clinical chemistry for many years. Serum proteins are separated into five fractions: albumin, ₁-globulin, ₂-globulin, ß-globulin, and -globulin. The clinical interpretation of electropherograms is based on the variation in the content of one or more of these five major fractions. In addition, detection and identification of paraproteins is important for the diagnosis of myeloma and monoclonal gammopathy of undetermined severity.

Established clinical electrophoretic methods use agarose or cellulose acetate as the separation base. However, handling these membranes and gels manually is labor intensive and technically demanding. Over the last few years, capillary zone electrophoresis (CZE)¹ has emerged as a powerful new tool for the rapid separation of various biopolymers, including proteins (1)(2)(3)(4)(5)(6). Separation by this technique depends on the electrophoretic mobility of the analyte and the electroosmotic flow of the bulk solution. Separations are fast and easily automated. The method uses UV detection at 214 nm for direct quantitation of proteins via the peptide bonds. Several recent reports have discussed and established the potential utility of serum proteins for clinical diagnostic applications (7)(8)(9)(10)(11). Separation patterns obtained by CZE resemble patterns from densitometric scans of cellulose acetate membrane electrophoresis (CAE) and agarose gel electrophoresis (AGE).

Recently, a dedicated automated system for the routine analysis of human serum proteins in clinical laboratories (Paragon 2000, Beckman Instruments) has become commercially available. High sample throughput is attained through the presence of seven fused-silica capillaries, which allow the simultaneous analysis of seven samples. In this study, the automated Paragon 2000 CZE system is evaluated and compared with conventional serum protein separations by CAE and AGE.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results and Discussion References

 
electrophoretic methods
For conventional AGE, we used the Paragon SPE kit (Beckman Instruments) according to the manufacturer's instructions. CAE was performed on Sepharose cellulose polyacetate electrophoresis strips for the Microzone System (Gelman Sciences). The gels were preequilibrated in diethylbarbital buffer, pH 8.6. We applied approximately 1 µL of each sample and carried out the electrophoresis at 220 V for 25 min in diethylbarbital buffer, pH 8.6. Proteins were stained by incubating the gels in Ponceau S (Analis) for 10 min. The individual fractions were quantified by densitometry (Beckman Appraise, Beckman Instruments). CZE was performed with the Paragon CZE^TM 2000 clinical capillary electrophoresis (Beckman Instruments). Proteins were measured by direct absorption at 214 nm through a small optical window in the capillary. The length of the fused-silica capillaries was 20 cm, and the i.d. was 20 µm. Instrument settings for serum protein electrophoresis were as follows: 1-min conditioning time, 1-s injection time, 4.3-min separation time, 0.5-min wash time, 0.5-min rinse time, 9000 V, and 24 °C. Sample dilution was 1:20. The running buffer was a borate buffer, and the capillary was rinsed between samples with <10 g/L NaOH cleaning solution (Beckman Instruments). The software used was Paragon, Ver. 1.08 (Beckman Instruments). The Paragon 2000 automatically selects pattern fractions, using an involved software package. The instrument attempts to find valleys to delimit fractions. In addition, delimit placement is based on the percentage position between two marker peaks (ranges and expected positions are company confidential information). The software uses the passed valley information to place delimits in the first valley (reading from -globulin to albumin) in the expected location range. If no valley exists within the range of the expected location, the delimit is placed at the center of the expected range. Manual editing of the cutoffs is possible. Such manual editing was necessary in 15–20% of the samples. In these samples, automatic delimitation did not separate the fractions adequately.

evaluation details
Reference intervals were established in accordance with NCCLS guideline C28-A (12). Protein electrophoresis and analysis of IgG, IgA, IgM, haptoglobin, serum complement component C3_c, albumin, transferrin, and ₁-acid glycoprotein (see below) were performed on all samples. Samples that were hemolytic, icteric, lipemic, that displayed a monoclonal component, or that showed an abnormal value for one of the specific proteins determined (e.g., 13 g/L IgA, haptoglobin <0.1 g/L) were excluded. Outlier exclusion was as described in NCCLS C28-A (12). When specimens were evaluated using these criteria, only 161 of the 200 specimens tested to establish reference intervals were acceptable.

Within-assay and total precision were calculated as described in NCCLS guideline EP5-T2 (13), using values obtained from two serum pools analyzed in duplicate during 20 days. Two aliquots of each pool were analyzed twice daily. The analyses were separated by a minimum of 2 h. The human serum pools were filtered through 8-µm (pore size) filters (Elkay Products), aliquoted, and stored at -20 °C until the assay. The protein concentrations of the pools were 71 and 76 g/L.

Method comparison was carried out by linear regression analysis, as described in NCCLS guideline EP9-A (14). Fifty patient specimens, including specimens from patients with multiple myeloma, were analyzed in duplicate over 8 days. Duplicate measurements were performed by reversing the order of the second aliquots.

Possible interfering agents that were investigated were hemoglobin, unconjugated bilirubin (Sigma-Aldrich), and lipemic samples that mimicked hemolyzed, icteric, and lipemic specimens, respectively. Six milligrams of bilirubin were dissolved in 400 µL of 0.1 mol/L NaOH containing 5 mmol/L EDTA. To this solution, 9 mL of serum pool and 400 µL of 0.1 mol/L HCl were added. Proportional amounts of HCl, NaOH, and EDTA were also added to another aliquot of the same serum pool. These sera were then mixed to make specimens with added bilirubin concentrations of 31.977 µmol/L (18.7 mg/L), 64.125 µmol/L (37.5 mg/L), 128.25 µmol/L (75 mg/L), 256.5 µmol/L (150 mg/L), and 513 µmol/L (300 mg/L). Specimens with no additional hemoglobin and with added hemoglobin concentrations of 9.3, 4.55, 2.35, 1.16, 0.58, and 0.29 g/L were prepared by a modification of the method of Glick et al. (15). Erythrocytes were lysed by adding distilled water. The samples then were placed in -20 °C for 12 h before centrifugation. To investigate the effects of lipids on serum protein electrophoresis, a normal serum sample was supplemented with a serum sample containing a high triglyceride concentration (10.26 mmol/L; 9.08 g/L). The high triglyceride sample was diluted with a saline solution (0.9% NaCl) to produce diluted lipemic samples with triglyceride concentrations of 0 (saline only), 8.20 mmol/L (7.26 g/L), 5.15 mmol/L (4.558 g/L), 4.10 mmol/L (3.63 g/L), 2.05 mmol/L (1.81 g/L), and 1.03 mmol/L (908 mg/L). Equal volumes of the diluted lipemic samples were mixed with equal volumes of the normal sample to produce test samples with final triglyceride concentrations of 5.13 mmol/L (4.54 g/L), 4.10 mmol/L (3.63 g/L), 3.07 mmol/L (2.72 g/L), 2.05 mmol/L (1.816 g/L), 1.03 mmol/L (908 mg/L), and 0.51 mmol/L (454 mg/L). The test samples were electrophoresed according to the procedures described above.

storage of samples and carryover
To study the effects of storage, CZE was performed on serum samples that had been kept at room temperature for 1, 2, 4, 8, 24, or 48 h and at 4 °C for 24 or 48 h after venipuncture.

To assess carryover in the -globulin fraction, specimens with high concentrations of -globulins and specimens with low concentrations of -globulins were selected. These two groups of samples were analyzed in the same capillary, running a low -globulin sample immediately after a high -globulin fraction. Midlevel specimens were prepared by combining equal volumes of the high- and low-concentration specimens.

other analyses
Serum proteins (IgG, IgA, IgM, haptoglobin, C3_c, albumin, transferrin, ₁-acid glycoprotein, -light chains, and -light chains) were quantitated by endpoint nephelometry, using a Behring BNA instrument (Behringwerke). All reagents and calibrators were from Behring (Behringwerke). Values were expressed according to the IFCC standardization based on CRM 470. Total protein concentration, glucose, complement-reactive protein, -glutamyltransferase, cholesterol, triglycerides, alanine amidotransferase, and total bilirubin were determined using Boehringer Mannheim reagent kits and applications on a Hitachi 747 automated analyzer (Boehringer Mannheim). Human ß-chorionic gonadotropin was determined by a Technicon Immuno-1 (Bayer), glycohemoglobin was detected with a Cobas Integra (Roche), hemoglobin was detected with a Celldyn (Abbott), and prothrombin time was determined with an ACL analyzer (Instrumental Laboratory).

determination of specific absorption coefficients
Albumin (Behring), ₁-antitrypsin, ₂-macroglobulin, C3, IgG (ICN Pharmaceuticals), haptoglobin, and transferrin (Sigma Chemical) were dissolved in borate (200 mmol/L, pH 7.4) at 50, 25, 12.5, 6.25, and 3.125 mg/L. In this concentration range, there was a linear relationship between the absorbance at 214 nm and the concentration. Specific absorption coefficients were calculated by averaging the values obtained at each concentration.

specimens
Specimens from healthy adults were obtained from blood donors (Red Cross, Belgium). The subjects were men and women aged 18 to 65 years. In addition, specimens from subjects affected by various pathological conditions (acute phase reaction, liver pathology, polyclonal gammopathy, diabetes, pregnancy, or nephrotic syndrome) were collected, aliquoted, and stored at -70 °C until analysis. The group with acute phase reaction (inflammatory syndrome) included patients with pneumonia, bronchitis, endocarditis, pancreatitis, pyelonephritis, meningitis, hepatitis B, HIV infection, or cystic fibrosis. Patients were selected if they had an elevated ₁-globulin and ₂-globulin fraction on CAE. Complement-reactive protein concentrations for this group ranged between 11 and 342 mg/L (median, 85 mg/L). The liver pathology group consisted of patients with cirrhosis (alcoholic and posthepatitic), cholecystitis, and choledocholithiasis. Within this group, total bilirubin concentrations were between 76.78 and 383.72 µmol/L (44.9–224.4 mg/L; median, 47.5 mg/L), -glutamyl transferase concentrations were between 87 and 506 U/L (median, 135 U/L), alanine amidotransferase concentrations were between 43 and 246 U/L (median, 55 U/L), and prothrombin time (%) was between 32% and 81% (median, 39%). The group with polyclonal gammopathy included patients with a tumor (non-Hodgkin's lymphoma and lung carcinoma), with HIV, with cystic fibrosis, or who had received a kidney transplant. These patients were selected on the basis of an elevated -globulin fraction on CAE. Patients included in the diabetes group had glycemia values between 11.27 and 20.42 mmol/L (2.03–3.68 g/L; median, 2.65 g/L) and glycohemoglobin concentrations between 8% and 11.5% (median, 9%). Pregnancy was confirmed by elevated ß-human chorionic gonadotropin values (between 15 556 and 96 100 U/L; median, 43 832 U/L). Patients with nephrotic syndrome all had total serum protein concentrations below 60 g/L and a strongly elevated ₂-globulin fraction.

racial distribution of subjects
All subjects included in the study were Caucasian.

statistical analysis
Statistical analysis was performed with the use of SAS, Ver. 6.11 (SAS Institute). The specific tests that were used are indicated in the text or in the legends to figures and tables.

	Results and Discussion

Top Abstract Introduction Materials and Methods Results and Discussion References

 
reference intervals
Reference intervals for the five electrophoretic serum protein fractions were determined by CAE, AGE, and CZE from a population of 82 men and 79 women. Median values and the central 95% of the distribution, expressed as fraction percentage and in concentration values (g/L), are presented in Table 1 . All distributions were within the normal range except for the ₂-globulin fraction concentration in women. Using Wilcoxon scores for the ₂-globulin fraction and t-test scores for all the other fractions, we observed statistically significant sex-related differences between the three electrophoretic methods for all fractions except the -globulin fraction (P <0.001 for albumin and P <0.05 for the ₁-globulin and ₂-globulin and ß-globulin fractions). When compared with AGE and CAE, substantially higher values for the ₁-globulin fraction were found with CZE, whereas lower values were found for the albumin (women), ₂-globulin, and ß-globulin fractions. AGE gave lower values for the -globulin fraction than did CZE and CAE. No statistically significant differences between AGE and CAE were observed for the albumin or the ₁-globulin and ₂-globulin fractions. Values for the ß-globulin fraction obtained by AGE were substantially higher than those obtained with CAE. The converse was found for the -globulin fraction. Serum concentrations of C3_c, transferrin, albumin, ₁-acid glycoprotein, haptoglobin, IgG, IgM, and IgA were determined immunochemically on the population that was used to establish the reference intervals. For men, the median and 95% reference intervals were as follows: for C3_c, 1.4 and 1.0–2.2 g/L; for transferrin, 2.7 and 2.0–3.5 g/L; for albumin, 45.2 and 39.1–52.5 g/L; for haptoglobin, 1.3 and 0.4–2.3 g/L; for IgG, 9.4 and 5.4–13.4 g/L; for IgM, 1.4 and 0.4–3.1 g/L; for IgA, 2.3 and 0.6–4.6 g/L; and for ₁-acid glycoprotein, 0.9 and 0.6–1.4 g/L . For women, the respective values were as follows: for C3_c, 1.3 and 0.9–1.9 g/L; for transferrin, 2.7 and 2.0–3.5 g/L; for albumin, 41.9 and 35.9–48.4 g/L; for haptoglobin, 1.3 and 0.4–2.3 g/L; for IgG, 9.4 and 5.4–13.4 g/L; for IgM, 1.4 and 0.4–3.1 g/L; for IgA, 1.6 and 0.7–3.7 g/L; and for ₁-acid glycoprotein, 0.7 and 0.5–1.3 g/L. Statistical analysis (Wilcoxon scores) showed significant sex-related differences for albumin, ₁-acid glycoprotein, C3_c, and IgA (P <0.001). These values of the serum proteins are superimposable on the recently proposed interim reference ranges for human serum proteins (16). This verifies the validity of the reference population.

We also compared the results of electrophoretic quantitation of albumin with those obtained by nephelometry and by the colorimetric assay on a Hitachi 747. Values for albumin (mean ± SD; n = 161) were 46.7 ± 3.2 g/L, 46.6 ± 3.2 g/L, 45.7 ± 3.3 g/L, 43.7 ± 3.9 g/L, and 42.9 ± 2.9 g/L when determined with CAE, AGE, CZE, nephelometry, and the Hitachi 747, respectively. Although values obtained with CZE were lower than those obtained with the two other electrophoretic methods, there was no statistically significant difference between the three methods (ANOVA and Scheffé test). Nor was there a difference between the nephelometric and colorimetric determination. However, there was a statistically significant difference (P <0.0001) between the electrophoretic methods and the nephelometric and colorimetric determinations.

linearity and protein quantitation
Concentrations of the various serum protein fractions may differ by a factor of 10–100. To verify that detector responses are equivalent across such a broad concentration range, we evaluated the linearity of the detection method. When a serum sample was diluted with increasing amounts (10%, 20%, 30%, 40%, 50%, 60%, and 70%) of saline, the results showed a linearity between 47.8 and 9.9 g/L for albumin (r = 0.99), between 4 and 0.8 g/L for ₁-globulin (r = 0.97), between 6.1 and 1.1 g/L for ₂-globulin (r = 0.99), between 6.4 and 1.2 g/L for ß-globulin (r = 0.99), and between 11.7 and 2.4 g/L for -globulin (r = 0.99). When a sample with an IgG monoclonal band (62 g/L by nephelometric determination) in the -globulin fraction was diluted with a normal serum sample, the results showed a linear relationship between fraction quantitation of the -globulin fraction and IgG concentrations between 13 and 46 g/L. The relationship was not linear at IgG concentrations >46 g/L. The results of the dilution of two other paraprotein samples showed a linear relationship up to 81 g/L (IgA paraprotein) and 118 g/L (IgG paraprotein).

In the Paragon 2000, protein is detected by measuring the absorbance. The wavelength used is 214 nm, corresponding to the high absorptivity of the polypeptide bonds of the protein molecules. The advantage of this method is that all proteins are quantitated. For example, the substantially higher values found for the ₁-globulin fraction with CZE when compared with CAE and AGE are related to the fact that both ₁-antitrypsin and ₁-acid glycoprotein are quantitated by direct absorption with CZE, whereas with conventional methods, the high sialic acid content of ₁-acid glycoprotein interferes with the binding of dyes used to quantitate the protein fractions (17). The direct absorption method, however, may be affected by the UV-active side chains found in phenylalanine, tryptophan, tyrosine, and histidine, which absorb light in the 240–280 nm range and which may interfere with the peptide bond absorbance at 214 nm. Therefore, we compared the abundance of the aromatic amino acids and histidine in the major serum proteins. The aromatic amino acids and histidine amounted to 9.9%, 11.7%, 9.8%, 11.0%, 11.2%, and 9% of the residues for albumin (585 amino acids), ₁-antitrypsin (394 amino acids), ₂-macroglobulin (1451 amino acids), transferrin (679 amino acids), haptoglobin 1S (329 amino acids), and complement C3 (1641 amino acids), respectively. This indicates that the overall aromatic content of these different proteins is comparable. In a second step, we determined the specific absorption coefficients (a_s) at 214 nm for a 1-cm light path of the most abundant serum proteins. The specific absorption coefficient (L·g^-1·cm^-1) was 13.61 for albumin, 17.96 for ₁-antitrypsin, 18.94 for ₂-macroglobulin, 17.42 for haptoglobin, 14.15 for transferrin, 10.41 for C3, and 13.32 for IgG. This shows that specific absorption coefficients differ between the various proteins and that these differences might lead to inaccuracy if concentrations of the specific proteins are calculated based on electrophoretic data. It should also be pointed out that substances such as drugs or contrast agents (e.g., Telebrix^® 35) that absorb at 214 nm can interfere with electropherograms (Weets I, Groven C, and Gerlo E, Free University Brussels, Belgium; communication in Medlab 97, Basel, CH).

precision
In a preliminary precision test, 20 aliquots from one serum sample were assayed in sequence by CZE, as proposed by NCCLS guideline EP5-T2 (13). The CVs were 0.76%, 2.83%, 2.26%, 0.96%, and 2.31% for albumin, ₁-globulin, ₂-globulin, ß-globulin, and -globulin, respectively. Total and within-run precision for protein electrophoresis by CZE was estimated according to NCCLS guideline EP5-T2 (13) for two samples. One sample was a pool prepared from normal sera, the other was serum from a patient with monoclonal gammopathy. The mean, SD, and CV were calculated for each fraction (Table 2 ). The precision decreased as the relative percentage of a fraction decreased. With one exception, CV values were <5%. This indicates that the Paragon 2000 instrument provides highly reproducible serum protein electrophoresis. The CV values obtained with CZE are clearly better than the published CV values for AGE (11) and CAE (18)(19). The CV values for CAE obtained in our laboratory were 1.6%, 8.8%, 6.3%, 7.9%, and 6.2% (n = 49, >25 days) for the albumin, ₁-globulin, ₂-globulin, ß-globulin, and -globulin fractions, respectively.

method comparison studies
To compare CZE with traditional AGE and CAE, 50 samples were analyzed on all three systems. The linear regression plots between AGE and CZE and between CAE and CZE are shown in Fig. 1 . Although good correlation between CZE and AGE had been published previously (8)(11), our results indicated only poor correlation between these two methods (Fig. 1 , upper panels). The r values were 0.54, 0.41, 0.53, 0.52, and 0.67 for the albumin, ₁-globulin, ₂-globulin, ß-globulin, and -globulin fractions, respectively. By contrast, good linear correlation was observed between CAE and CZE for albumin (r = 0.94), ₁-globulin (r = 0.85), ₂-globulin (r = 0.94), and -globulins (r = 0.97; Fig. 1 ). For the ß-globulin fraction, r = 0.67. Only poor linear correlation was found between CAE and AGE, with r values <0.7 (data not shown).

View larger version (27K): [in this window] [in a new window]   Figure 1. Linear correlation between AGE and CZE and between CAE and CZE. Linear correlation was evaluated on 50 samples, as described in Materials and Methods. Samples were from hospitalized patients and included 28 samples obtained from females (ages 7–97 years) and 22 samples obtained from males (ages 5–79 years). Three samples displayed a monoclonal peak. The upper panels show the correlation between AGE (x-axis) and CZE (y-axis) for the albumin, 1-globulin, 2-globulin, ß-globulin, and -globulin fractions. The lower panels show the correlation between CAE (x-axis) and CZE (y-axis) for the albumin, 1-globulin, 2-globulin, ß-globulin, and -globulin fractions.

stability after venipuncture and carryover
The mean ± SD values for five serum protein fractions determined by CZE at various time points after venipuncture are shown in Table 3 . Keeping samples for 24 or 48 h at room temperature resulted in a statistically significant decrease of the ₂-globulin fraction. No changes were noted when the samples were stored at 4 °C.

To evaluate carryover on the Paragon 2000, samples with a low -globulin concentration (12.8% ± 0.5% of total protein; n = 8) were run in the same capillary immediately after samples containing a high (35.4% ± 18.5% of total protein; n = 8) or a medium (24.7% ± 10.5% of total protein; n = 8) concentration of -globulin. Similarly, samples with a medium concentration of -globulin were run in the same capillary immediately after samples with a high -globulin concentration. The concentration of -globulin in the low concentration samples analyzed after medium concentration samples was 12.8% ± 0.6% (n = 8), compared with 12.7 ± 0.7% (n = 8) for the low concentration samples analyzed after high concentration samples and 24.9% ± 10.6% (n = 8) for the medium concentration samples analyzed after high concentration samples. Statistically, neither of these values was significantly different (Kruskal–Wallis test) from the control group, indicating an absence of carryover.

interferences
We investigated the effects of hemoglobin, lipids, and unconjugated bilirubin on AGE, CAE, and CZE by adding various concentrations of human hemolysate, lipid, or bilirubin to normal serum.

Addition of unconjugated bilirubin did not affect the fractionation results in any of the three methods (data not shown).

Increasing concentrations of hemoglobin resulted in a progressive decrease of the albumin fraction (Fig. 2 , lower panel) and a gradual but substantial increase of the ß-globulin fraction (Fig. 2 , upper panel) in all methods. Only minor effects were seen on the ₂-globulin fraction; the concentration of ₂-globulin measured by CAE, AGE, and CZE was 10.3%, 8.9%, and 8.6% before the addition of 9.3 g/L hemoglobin and 9.1%, 12%, and 9.8% after the addition, respectively.

View larger version (16K): [in this window] [in a new window]   Figure 2. Effect of hemolysis on electrophoresis. Normal serum was supplemented with increasing concentrations of human hemolysate, as described in Materials and Methods, and CAE (), AGE (), and CZE () were performed. The top panel illustrates the effect of increasing amounts of hemoglobin on the ß-fraction. The bottom panel shows the effect on the albumin fraction.

We evaluated 12 samples with an elevated triglyceride concentration ranging between 3.92 and 10.26 mmol/L (3.47–9.08 g/L). Nine of the 12 samples displayed an abnormal morphology of the ₂-globulin fraction, including the presence of a small interference peak. To investigate the effects of lipids on capillary electrophoresis further, we supplemented a normal serum with a selected serum sample that contained a high triglyceride concentration (10.26 mmol/L; 9.08 g/L) (see Materials and Methods). The abnormal morphology of the ₂-globulin fraction was apparent in the test samples with added triglyceride concentrations of 5.13 mmol/L (4.54 g/L) (Fig. 3 A), 4.10 mmol/L (3.63 g/L), 3.07 mmol/L (2.72 g/L) (Fig. 3B ), 2.05 mmol/L (1.816 g/L) (Fig. 3C ), and 1.03 mmol/L (908 mg/L). The interference peak diminished with decreasing triglyceride concentrations and was absent in the test samples to which 0.51 mmol/L (454 mg/L; Fig. 3D ) and no triglyceride was added.

View larger version (14K): [in this window] [in a new window]   Figure 3. Effect of lipids on electrophoresis. Normal serum was supplemented with a serum sample containing a high triglyceride concentration (9.08 g/L) (see Materials and Methods) to obtain test samples with decreasing triglyceride concentrations. The figures show the electropherograms of test samples containing 4.54 g/L (A), 2.724 g/L (B), 1.81 g/L (C), and 454 mg/L (D) added triglycerides. The arrows indicate the abnormal morphology in the 2-globulin fraction.

Fibrinogen due to delayed coagulation in serum collected in plastic tubes or in serum collected from patients treated with heparin-type anticoagulant is seen as a peak in the -globulin zone with CAE (Fig. 4 A) and AGE (Fig. 4B ); the fibrinogen peak is indicated by an arrow. In contrast, fibrinogen cannot be discerned on the electropherogram (Fig. 4C ) from the CZE. Similarly, with Paragon 2000, no fibrinogen peak can be found in plasma samples or in serum samples supplemented with 10 g/L fibrinogen (Analis–Beckman, personal communication). The reason underlying this observation is not known to the authors. Fibrinogen is soluble in the sample and running buffers (Analis–Beckman, personal communication). Therefore, the absence of a fibrinogen peak in the CZE electropherogram is not likely to be from the precipitation of fibrinogen. This feature may prove to be an advantage of CZE over CAE and AGE because, in the latter two techniques, fibrinogen can simulate a paraprotein and/or interfere with paraprotein detection.

View larger version (23K): [in this window] [in a new window]   Figure 4. Comparison of electrophoretic patterns in various conditions. A–C, comparisons of electrophoretic patterns in a patient receiving anticoagulant therapy. Electropherograms were obtained with CAE (A), AGE (B), or CZE (C). The arrows in A and B indicate the position of fibrinogen. D–F, comparison of electrophoretic patterns in a patient with nephrotic syndrome. Electropherograms were obtained with CAE (D), AGE (E), or CZE (F). The arrows in D and E indicate the position of 2-macroglobulin. G–I, comparison of electrophoretic patterns in a patient with an inflammatory syndrome. Electropherograms were obtained with CAE (G), AGE (H), and CZE (I). The arrows in G and H indicate the position of haptoglobin.

method comparison of electrophoretic patterns
We first compared the protein profiles of 524 specimens from hospitalized adult patients by routine electrophoresis using CAE and CZE. For each specimen, results were compared with the respective reference intervals established for the method. When the result of a protein fraction exceeded the reference interval, it was marked as or -, or as or -- when it exceeded the reference interval by more than 1 SD. Under these criteria, classification by the two electrophoretic methods was identical for 202 specimens. In 322 specimens, differences between the two methods were found for one or more fractions. For the albumin fraction and the -globulin fraction, higher values (i.e., vs normal, vs , vs normal, normal vs -, normal vs --, or - vs --) were found with CAE (66 specimens) than with CZE (36 specimens). Higher values were found with CZE than with CAE in 12 specimens for the albumin and 5 specimens for -globulin fraction. For the ß-globulin fraction, CZE gave higher values than CAE in 82 cases, whereas the converse was true in only 18 cases. No systematic differences between the two methods were observed for the ₁-globulin and ₂-globulin fractions.

To evaluate whether the clinical interpretation was concordant between CAE and CZE, a blind interpretation of 524 electropherograms was performed, using the following classifications: normal, acute phase reaction, chronic inflammation, hypoalbuminemia, hypogammaglobulinemia, and polyclonal elevation of the -globulins. Acute phase reaction was defined by an elevation of both -globulin fractions, whereas chronic inflammation was defined by elevated ₁-globulin, ₂-globulin, and -globulin fractions and decreased albumin. Discordant interpretation was found in 67 cases. In 39 cases, acute phase reaction was suggested by CAE but not by CZE. With the latter technique, an elevation of only one fraction (mostly ₂-globulin) was seen in most of the cases. The converse situation was found in 13 cases. The 15 other cases of discordance between CAE and CZE were as follows. CZE detected hypogammaglobulinemia (four cases), hypoalbuminemia (six cases), or a polyclonal increase of the -globulins (one case), whereas no changes were detected by CAE. The converse situation was found in four cases, i.e., normal CZE electropherograms in the presence of hypogammaglobulinemia (one case), hypoalbuminemia (two cases), and polyclonal increase of the -globulins (one case).

We next specifically examined whether the characteristic abnormalities detected by CAE and AGE in the sera of patients with certain pathologies, e.g., acute phase reaction (inflammatory syndrome) or nephrotic syndrome, would be equally detected by CZE. Therefore, several patient groups were analyzed with CAE, AGE, and CZE. Patients were divided by gender, and values were statistically compared with the sex-specific and method-specific reference intervals. Because reference values differ depending on the method, no direct comparison between values obtained by the three methods was performed for each pathological group. Specific proteins, including IgA, IgM, haptoglobin, C3_c, albumin, transferrin, and ₁-acid glycoprotein, were determined by nephelometry. The results are presented in Table 4 .

The characteristic electrophoretic changes for the acute phase reaction (inflammatory syndrome) are depressed albumin and increased ₁-globulin, ₂-globulin, and -globulin fractions. These changes were found in both male and female patients by all three electrophoretic methods. For each method, statistically significant deviations from the method-specific reference values were observed for the albumin ₁-globulin, ₂-globulin, and -globulin fractions. CZE additionally detected an increased ß-globulin fraction. Values obtained by nephelometric analysis of the samples from the male patients were compared with reference intervals (see above) and found to be statistically significant (Wilcoxon scores) in the decreased concentrations of transferrin (1.68 g/L, P <0.001) and albumin (30.18 g/L, P <0.001) and in the increased concentrations of haptoglobin (2.53 g/L, P <0.05), ₁-acid glycoprotein (1.85 g/L, P <0.01), IgG (16.61 g/L, P <0.001), and IgA (4.24 g/L, P <0.01). Similarly, values from the female patients showed statistically significant (Wilcoxon scores) decreased concentrations of transferrin (2.19 g/L, P <0.01) and albumin (32.75 g/L, P <0.001) and increased concentrations of haptoglobin (2.51 g/L, P <0.05), ₁-acid glycoprotein (1.65 g/L, P <0.001), IgA (4.20 g/L, P <0.05), and C3_c (4.2 g/L, P <0.01). These changes in the protein profile are typical for an acute phase reaction (inflammatory syndrome) and verify the validity of the patient groups.

In patients classified as having polyclonal gammopathy, all three electrophoretic methods showed statistically significant increased concentrations of ₁-globulin and -globulin fractions combined with a decreased albumin fraction. An elevated ₂-globulin fraction was found with CZE but not with CAE or AGE. Nephelometric analysis of this patient group showed a statistically significant reduction of albumin (37.88 g/L, P <0.001) and transferrin (2.05 g/L, P <0.01) concentrations and increased values for ₁-acid glycoprotein (1.51, P <0.001), IgG (22.27 g/L, P <0.001), IgA (4.78 g/L, P <0.05), and IgM (2.64 g/L, P <0.01). The reduced transferrin concentration found by nephelometry was not reflected in any of the three electrophoretic methods.

The typical pattern of decreased albumin and -globulin fractions with a markedly enhanced ₂-globulin fraction in sera of patients with nephrotic syndrome was detected by each of the three electrophoresis methods. An elevated ₁-globulin fraction was detected by CZE but not by CAE and AGE. In this patient group, nephelometry confirmed statistically (Wilcoxon scores) reduced concentrations of albumin (20.96 g/L, P <0.01), transferrin (1.97 g/L, P <0.001), and IgG (2.45 g/L, P <0.001). Elevation of the ₂-globulin fraction in the serum protein profile is seen in patients with nephrotic syndrome (Fig. 4 , D–F) and in patients with an acute phase reaction (inflammatory syndrome; Fig. 4 , G–I). Nephrotic syndrome gives rise to a high ₂-macroglobulin concentration associated with hypoproteinemia. Severe inflammatory syndrome gives rise to a very high haptoglobin concentration, concomitant with an elevated ₁-globulin concentration. Due to the more anodal migration of ₂-macroglobulin, increased ₂-macroglobulin concentrations (Fig. 4 , D and E) can be differentiated from increased haptoglobin concentrations by CAE and AGE (Fig. 4 , G and H). In contrast, CZE did not distinguish between haptoglobin and ₂-macroglobulin. The CZE electropherogram of a patient with elevated ₂-macroglobulin is shown in Fig. 4F , and the CZE electropherogram of a patient with an elevated haptoglobin is shown in Fig. 4I .

Specimens from diabetic patients displayed an increased ₂-globulin concentration in all electrophoretic methods. An elevation of the ₁-globulin fraction in male patients was found with CZE but not with CAE and AGE.

Serum from patients with liver pathology contained a reduced albumin fraction and elevated ₁-globulin and -globulin fractions in all electrophoretic methods. AGE also showed an increased ß-globulin fraction. Evaluation of the specific serum proteins substantiated reduced albumin (29.8 g/L; P <0.01, Wilcoxon score) and transferrin (1.72 g/L; P <0.001, Wilcoxon score) concentrations. The increase in the concentration of the immunoglobulins was not statistically significant. The main electrophoretic features of pregnancy were decreased albumin and increased ₁-globulin, ₂-globulin, and ß-globulin fractions. These changes were observed with all electrophoretic methods studied.

As illustrated in Fig. 5 , CZE easily detected bisalbuminemia (Fig. 5A ) and ₁-antitrypsin deficiency (Fig. 5B ).

View larger version (10K): [in this window] [in a new window]   Figure 5. CZE electropherograms of samples from a patient with bisalbuminemia (A) and a patient with ZZ phenotype 1-antitrypsin deficiency (B). The arrow in B indicates the position of the 1-globulin fraction.

In conclusion, electrophoretic patterns and clinical information obtained by CZE are comparable with the patterns and the clinical information obtained by classical AGE or CAE. The results related to paraprotein detection are represented and discussed in the accompanying paper (16).

	Acknowledgments

 
We acknowledge M. Artoos, H. Raveschot, and M. Wuyts for their expert technical assistance. We thank G. Mariën and E. Stevens for their contribution in the generation of the nephelometric data. We also thank Analis, Belgium, for providing the instrumentation and reagents for capillary electrophoresis and AGE.

	Footnotes

 
Central Clinical Laboratory, Department of Clinical Pathology, University Hospital of Leuven, Kapucijnenvoer 33, B-3000 Leuven, Belgium.

¹ Nonstandard abbreviations: CZE, capillary zone electrophoresis; AGE, agarose gel electrophoresis; and CAE, cellulose acetate electrophoresis.

	References

Top Abstract Introduction Materials and Methods Results and Discussion References

 

1. Gordon MJ, Huang X, Pentoney SL, Zare RN. Capillary electrophoresis. Science 1988;242:224-228. [Abstract/Free Full Text]
2. Mazzeo JR, Krull JR. Capillary electrophoresis: the promise and the practice. Nature 1992;357:92-94. [Medline] [Order article via Infotrieve]
3. Landers JP. Capillary electrophoresis: pioneering new approaches for biomolecular analysis. Trends Biochem Sci 1993;18:409-414. [Web of Science][Medline] [Order article via Infotrieve]
4. Reif OW, Lausch R, Freitag R. High-performance capillary electrophoresis of human serum and plasma proteins. Adv Chromatogr 1994;34:1-56. [Medline] [Order article via Infotrieve]
5. Dolnik V. Capillary zone electrophoresis of serum proteins: study of separation variables. J Chromatogr A 1995;709:99-110. [Web of Science][Medline] [Order article via Infotrieve]
6. Lehmann R, Liebich H, Grübler G, Voelter W. Capillary electrophoresis of human serum proteins and apolipoproteins. Electrophoresis 1995;16:998-1001. [Web of Science][Medline] [Order article via Infotrieve]
7. Chen FTA, Liu CM, Hsieh Y-Z, Sternberg JS. Capillary electrophoresis–a new clinical tool. Clin Chem 1991;37:14-19. [Abstract/Free Full Text]
8. Kim JW, Park JH, Park JW, Doh HJ, Heo GS, Lee KJ. Quantitative analysis of serum proteins separated by capillary electrophoresis. Clin Chem 1993;33:689-692. [Abstract/Free Full Text]
9. Chen FTA, Sternberg JC. Characterization of proteins by capillary electrophoresis in fused-silica columns: review on serum protein analysis and application to immunoassays. Electrophoresis 1994;15:13-21. [Web of Science][Medline] [Order article via Infotrieve]
10. Landers JP. Clinical capillary electrophoresis. Clin Chem 1995;41:495-509. [Abstract/Free Full Text]
11. Wijnen PAHM, Dieijen-Visser MP. Capillary electrophoresis of serum proteins. Reproducibility, comparison with agarose gel electrophoresis and a review of the literature. Eur J Clin Chem Clin Biochem 1996;34:535-545. [Web of Science][Medline] [Order article via Infotrieve]
12. NCCLS. How to define and determine reference intervals in the clinical laboratory: approved guideline. NCCLS Document C28-A. Villanova, PA: NCCLS, 1992:.
13. NCCLS. Evaluation of precision performance of clinical chemistry devices, 2nd ed: tentative guideline. NCCLS Document EP5–T2. Villanova, PA: NCCLS, 1992;12(4)..
14. NCCLS. Method comparison and bias estimation using patient samples: approved guideline. NCCLS Document EP9-A. Wayne, PA: NCCLS, 1995;..
15. Glick MR, Ryder KW, Jackson SA. Graphical comparisons of interferences in clinical chemistry instrumentation. Clin Chem 1986;32:470-475. [Abstract/Free Full Text]
16. Bossuyt X, Bogaerts A, Schiettekatte G, Blanckaert N. Detection and classification of paraproteins by capillary immunofixation/subtraction. Clin Chem 1998;44:760-764. [Abstract/Free Full Text]
17. Dati F, Schumann G, Thomas L, et al. Consensus of a group of professional societies and diagnostic companies on guidelines for interim reference ranges for 14 proteins in serum based on the standardization against the IFCC/BCR/CAP reference material (CRM 470). Eur J Clin Chem Clin Biochem 1996;34:517-520. [Web of Science][Medline] [Order article via Infotrieve]
18. Keren DF. High-resolution electrophoresis and immunofixation techniques and interpretation, 2nd ed 1994:58 Butterworth Stoneham, MA. .
19. Baars JD, Lombarts AJ. Imprecision of protein electrophoresis. Clin Chem 1986;32:1425-1426. [Web of Science][Medline] [Order article via Infotrieve]
20. Kahn SN, Strony LP. Imprecision of quantification of serum protein fractions by electrophoresis on cellulose acetate. Clin Chem 1986;32:356-357. [Abstract/Free Full Text]