Differences in protein fractions of avian plasma among three commercial electrophoresis systems.
Key words: protein electrophoresis, avian, African grey parrots, Psittacus erithacus
Protein electrophoresis (EPH) has been found to be a valuable tool in human and veterinary clinical pathology. (1-3) Nearly 40 years ago, transitions in techniques from paper electrophoresis to polyacrylamide medium and cellulose acetate membranes provided the first opportunity to increase the precision of this technique. (4,5) In more recent years, several commercial vendors have produced semiautomated systems using thin-layer agarose gels. These systems have been widely used in both human and veterinary clinical laboratories, and a good precision has been generally found among these reagents when analyzing human specimens.
In the United States, the Beckman Paragon system (Beckman Coulter, Fullerton, CA, USA) was used to establish reference intervals on several species of psittaciform birds, and this system was generally in wide use in clinical laboratories offering testing for these species. (6-8) In mid-2009, Beckman discontinued this product line, leaving 2 primary commercial alternatives using agarose gels: Helena (Helena Laboratories, Beaumont, TX, USA) and Sebia (Sebia Inc, Norcross, GA, USA). Our laboratory conducted extensive testing of both systems and observed apparent significant differences between the Sebia and Beckman systems, especially with specimens from psittaciform species. The purpose of the current study was to describe these differences as well as those between the Beckman and Helena systems by using traditional method comparison techniques. (9)
Materials and Methods
Heparinized plasma specimens were obtained from 40 African grey parrots (Psittacus erithacus) from an aviary. Specimens were aliquoted in freezer-type tubes (O-ring enclosures) and frozen at -70[degrees]C (-94[degrees]F). All samples were free of visible hemolysis and lipemia. Total protein was determined by the biuret method with an Ortho Vitros 250 analyzer (Ortho Clinical Diagnostics, Rochester, NY, USA). The total protein concentration was used to calculate the concentrations of the individual protein fractions in the EPH assay.
The EPH assays were run using the Beckman Paragon system with SPEP-II gels (Beckman Coulter), Sebia Hydrasys system with Hydragel 7 Protein(e) gels (Sebia Inc), and the Helena SPIFE 3000 system with Split Beta gels (Helena Laboratories). The Beckman and Helena studies were conducted at the University of Miami (Miami, FL, USA), and the Sebia study was conducted at the ALX Laboratory (New York, NY, USA) after shipment of specimens on dry ice. For both test sites, the specimens only underwent a single freeze-thaw before analysis. A minimum of 8-10 samples were run per day until the completion of the study. No fibrin clots or other possible artifacts of storage were observed with the specimens. At the time of this study, the University of Miami held more than 15 years experience with the Beckman system and 1 year of experience with the Helena system. The ALX Laboratory had 2 years of experience with the Sebia system.
Beckman and Helena results were obtained from densitometry tracings with fraction delimitation to create a fraction pattern as previously described for the Beckman system. (7) Sebia results were obtained with fraction delimitation to create a fraction pattern as performed by Roman et al. (10,11) Delimits are automatically placed by the densitometry or scanning software. Any necessary changes were completed by experienced staff members to create the fraction patterns. These changes were completed while performing a visual interpretation of the stained gel. All the EPH systems produce results that express quantitation of fraction percentage to the tenths decimal. This percentage was used to calculate the concentration of each fraction by using the total protein concentration. Traditionally, the final fraction concentration is expressed to the hundredths decimal using grams per deciliter.
[FIGURE 1 OMITTED]
Pooled stock from fresh samples from healthy psittaciform birds was prepared, aliquoted, and stored for use. Within-run analysis included a minimum of 8 aliquots. Day-to-day analysis was performed over a minimum of 10 days. For all coefficient of variation (CV) analyses, calculations were performed using absolute fraction concentrations. The within-run and day-to-day results were combined to calculate the total CV for each fraction. The combined, inherent imprecision of both instruments was calculated as previously described. (9) If >95% of the differences were within this expected combined imprecision, this would indicate that the results between the EPH systems are not different. (9)
All analyses were conducted using SAS software (SAS, Cary, NC, USA) and MedCalc (MedCalc Software, Mariakerke, Belgium). The data comparison was conducted using Pearson correlation and Passing-Bablok regression analysis. Bland-Altman plots were used to detect bias. Significance was set at P < .05. Pearson correlation (r) coefficients were considered excellent (0.9-1.0), high (0.70-0.89), good (0.50-0.69), low (0.30-0.49), and poor (<0.29). Passing-Bablok analysis was used to define the presence of constant and proportional error. If the 95% confidence interval (CI) for the slope did not include 1, this was considered reflective of proportional error. If the 95% CI for the [gamma]-intercept did not include 0, this was considered evidence of constant error.
[FIGURE 2 OMITTED]
[FIGURE 3 OMITTED]
The same 40 African grey plasma specimens were analyzed on all 3 EPH systems. Consistent differences were observed in the electrophoretogram patterns. As shown in Figure 1, whereas similar fraction patterns were found between Beckman and Helena systems, results from the Sebia system demonstrated a decreased prealbumin fraction and increased [[alpha].sub.1] fraction. This difference in protein patterns was reflected in the mean values of the EPH fractions (Table 1).
An imprecision study was conducted (Table 2). Both day-to-day and within-run analyses were conducted and reported as total CV. The imprecision data are different among the systems. Acceptable differences, based on the combined system imprecision data, were used in the Bland-Altman plots (Figs 2 through 7).
[FIGURE 4 OMITTED]
Pearson correlations and Passing-Bablok analyses were conducted. For the comparison of the Beckman and Helena systems, high or excellent correlation was observed, with the exception of a poor correlation for [beta] globulins (Table 3). Constant error was observed for prealbumin, [[alpha].sub.2], and [beta] globulin fractions, and proportional error was observed for the albumin fraction. For the comparison of the Beckman and Sebia systems, the prealbumin, [[alpha].sub.1], and [beta] globulin fractions were observed to have poor or low correlations (Table 4). The [[alpha].sub.2] fraction had a good correlation; the albumin and y globulin fractions had a high correlation. Prealbumin, [beta] globulin, and [gamma] globulin comparisons demonstrated the presence of a constant error. Proportional error was observed with the prealbumin, [[alpha].sub.1], [beta] globulin, and [gamma] globulin fractions.
[FIGURE 5 OMITTED]
Bland-Altman plots further demonstrate the presence of proportional or constant error (Figs 2 through 7). Combined imprecision was also calculated and placed in these plots as presented with the dotted lines. The percentage of difference between the Beckman and Helena systems that were within the imprecision limits ranged from 25% to 80% (Table 5). For the Beckman and Sebia systems, the percentage ranged from 0% to 72.5%.
A method comparison study between the Beckman and the newly implemented Helena and Sebia EPH systems was undertaken using recommended assessments and statistical analyses. (9,12) Plasma samples from African grey parrots were selected based on availability and because they are common species from which samples are submitted for EPH fraction analysis. In addition, per previous work with the Beckman system, these specimens provide a consistent 6-fraction electrophoretogram, which this laboratory has studied extensively. (7) Of note, fraction differences among the analyzers described within the current study have also been observed across many avian species, especially those with a defined prealbumin fraction (data not shown). Based on recommendations for method comparison studies, 40 specimens were examined. (9,12,13) Two laboratories conducted the analyses of these specimens. All laboratories and technicians had a minimum of 1 year of experience with these systems, which represents a minimum of 5000 avian specimens reviewed at the time of this study. Each system had the same specimen requirements, and calibration and preventative maintenance were performed per manufacturers' recommendations.
[FIGURE 6 OMITTED]
[FIGURE 7 OMITTED]
Precision studies demonstrated differences between the EPH systems. Total CVs were similar between the Beckman and Helena systems, with the exception of an elevated CV for prealbumin in the Helena system. Using human samples, very similar CVs were observed with the exception of the prealbumin fraction, which is not commonly identified in human samples (Helena, technical insert, protein fraction CV range: 2.0%-10.9%). In the comparison of the Beckman and Sebia systems, an increase in the prealbumin CV but a marked decrease in the CV for the [[alpha].sub.1] fraction was observed. These changes are consistent with the differences in the electrophoretograms between these systems. That is, good fraction precision has been associated with the presence of a better visually defined protein fraction.; For example, contrast the near lack of a defined [[alpha].sub.1] fraction in the Beckman and Helena systems versus the fraction present using the Sebia system (Fig 1). Using human samples, a CV range of 0.8%-4.6% was observed (Sebia, technical insert). The poorly defined prealbumin and [[alpha].sub.2] fractions exceed this CV range when analyzing avian specimens. Thus, as the values of fractions depend on the fraction delimitation, variable precision in "cutting" of the fractions may result in a relative error in adjacent fractions.
Initial statistical analyses of the data included the use of the Pearson correlation coefficient. This technique assesses association but not agreement. Consistent levels of correlation were not observed among all fractions of the EPH systems. Results from the Helena system had high or excellent correlation for all fractions, except [beta] globulins, when compared with the Beckman system. In the comparison of the Sebia and Beckman systems, only the albumin and y globulin exhibited a high correlation. As no correlation value was greater than 0.975, further analysis was conducted using the Passing-Bablok regression analysis. Only the [[alpha].sub.1] and [gamma] globulin fractions were found to be free of constant or proportional error when comparing the Beckman and Helena systems. A marked difference was present between prealbumin results using the Beckman and Sebia systems such that 95% CI could not be calculated. Only the albumin and [[alpha].sub.2] globulin fractions were free of constant or proportional error.
To judge the acceptability of the new Helena and Beckman methods, Bland-Altman plots were used. As data on biological variation has not been widely defined on psittaciform species, instead of basing acceptability on total error (or analytical quality specifications), we used a combined inherent imprecision calculated from the total CV data. (9) To be an acceptable method by this statistical analysis, 95% of the differences between the methods should be within the range of total imprecision. Although some fractions ranged from 72.5%-80%, many fractions were <50%. This indicates that neither the Helena nor the Sebia systems are identical to the Beckman system.
Potential weaknesses of this current study include sample selection and handling. As extensive differences were examined with this pool of specimens from clinically healthy birds, additional studies were not undertaken to increase that sample size. Increases in protein fractions, which are often found in birds with clinical abnormalities, possibly result in a more precise definition of protein fractions. To conduct specimen analysis at the 2 laboratory facilities, all the samples were exposed to a single freeze-thaw cycle. Given previous stability studies, we do not believe that doing so would have significantly affected the current data. (7)
The basis of the differences among the protein fractions in avian species is especially interesting given that all 3 of these systems have been commonly used in human clinical pathology laboratories and appear to provide similar results on specimens from mammals (C. C., personal observation). Differences in fraction migration by avian species has been previously reported. (7,11,14) Notably, Roman et al (10) recently reported the presence of apolipoprotein A-1 in the large [[alpha].sub.1] globulin fraction of Amazon parrots (Amazona species) using the Sebia system. In those same specimens, there was no evidence of a prealbumin fraction, which has been described using the Beckman system. (7) Thus, these fractions may possibly migrate differently in different EPH systems. In addition, protein stains and staining time may also result in a difference in fraction quantitation. Among the 3 systems, methodological differences are present in agarose concentration, running buffer, voltage, running time, protein stains, and densitometer. The Beckman system used a 1% agarose gel, barbital running buffer with a pH of 8.6, a run time of 37 minutes at 100 volts, and Coomassie blue stain. The Sebia system uses 0.8% agarose gels, barbital running buffer at pH 9.2, a run time of 7 minutes at 10 watts, and amidoblack stain. The Helena system runs 10 minutes at 650 volts and uses a barbital buffer and acid blue stain. The other reagent information is proprietary. Differences in agarose concentration, voltage/run time, and buffer pH may affect fraction migration. As the protein stains have been optimized for the detection of human proteins, staining differences may be present among mammalian and avian proteins. In reference to all these possible etiologies, our laboratory extensively varied voltage, run time, buffers, and stains using the Sebia system and were unable to produce fraction migrations similar to either the Beckman or Helena systems (data not shown).
Because a laboratory technician must be trained to consistently place fraction delimits, electrophoresis as a technique does have an inherent imprecision. (4,7,15) In a study in which multiple aliquots of avian specimens were blinded and shipped to commercial laboratories, variable reliability of fraction quantitation was reported. (16) Although this may have been related to sample handling and a poor representation of the working range of the protein fractions, the recommended use of a CV analysis rather than a correlation coefficient demonstrated a variation mostly consistent with Beckman reagent guidelines using human samples. (7,16,17) In both previous Beckman studies, as well as current Helena and Sebia data, results show that avian samples do present special challenges in making consistent fraction delimitation, and laboratories that conduct EPH analyses should work within a defined, structured protocol with these systems.
Given the dissimilarities demonstrated in our study, clinicians should be careful when comparing patient results obtained before 2009 to those obtained with the new EPH systems. Additionally, they should identify and have continuity of use with the laboratory they select for avian EPH analyses because both Sebia and Helena systems are now in use in the different major clinical laboratories that offer specialized avian diagnostic testing in the United States. Clinical pathologists at these laboratories should be aware of these method differences and prepare new reference intervals for avian species. An examination of avian fraction migration using the Sebia system has been partially completed. (10,11) Studies are currently planned to further define the protein fractions in the Helena system, as we previously described with the Beckman system and, in particular, to identify the composition of the prealbumin fraction.
(1.) Alper CA. Plasma protein measurements as a diagnostic aid. N Engl J Meal. 1974;291(6):287-290.
(2.) Dimpopullus GT. Plasma proteins in health and disease. Ann N Y Acad Sci. 1961;94:1 8.
(3.) Kaneko JJ. Serum proteins and the dysproteinemias. In: Kaneko JJ, Harvey JW, Bruss ML, eds. Clinical Biochemistry of Domestic Animals. 5th ed. San Diego, CA: Academic Press; 1997:117-138.
(4.) Kahn SN, Strony LP. Imprecision of quantification of serum protein fractions by electrophoresis on cellulose acetate. Clin Chem. 1986;32(2):356-357.
(5.) Hyslop NS. Application of an improved system of electrophoresis in acrylamide gel to studies on the sera of different species. J Clin Pathol. 1972;25(6): 508-511.
(6.) Cray C, Tatum LM. Applications of protein electrophoresis in avian diagnostic testing. J Avian Med Surg. 1998;12(1):4-10.
(7.) Cray C, Rodriguez M, Zaias J. Protein electrophoresis of psittacine plasma. Vet Clin Pathol. 2007;36(1):64-72.
(8.) Werner LL, Reavill DR. The diagnostic utility of serum protein electrophoresis. Vet Clin North Am Exot Anim Pract. 1999;2(3):651-662.
(9.) Jensen AL, Kjelgaard-Hansen M. Method comparison in the clinical laboratory. Vet Clin Pathol. 2006;35(3):276-286.
(10.) Roman Y, Bed'hom B, Guillot A, et al. Identification of apolipoprotein A-I in the alpha-globulin fraction of avian plasma. Vet Clin Pathol. 2009; 38(2):206-212.
(11.) Roman Y, Levrier J, Ordonneau D, et al. Location of the fibrinogen and albumin fractions in plasma protein electrophoresis agarose gels in five taxonomically distinct bird species. Revue Med Vet. 2009;160(3): 160-165.
(12.) Koch DD, Peters T. Selection and evaluation of methods--with an introduction to statistical techniques. In: Burtis CA, Ashwood ER, eds. Tietz Fundamentals of Clinical Chemistry. 5th ed. Philadelphia, PA: WB Saunders; 2001:320-335.
(13.) Bellamy JEC, Olexson DW. Evaluating laboratory procedures. In: Quality Assurance Handbook for Veterinary Laboratories. Ames: Iowa State University Press; 2000:61-77.
(14.) Archer FJ, Battison AL. Differences in electrophoresis patterns between plasma albumins of the cockatiel (Nymphicus hollandicus) and the chicken (Gallus gallus domesticus). Avian Pathol. 1997; 26(4):865-870.
(15.) Laurell CB. Electrophoresis, specific protein assays, or both in measurement of plasma proteins? Clin Chem. 1973;19(1):99 102.
(16.) Rosenthal KL, Johnston MS, Shofer FS. Assessment of the reliability of plasma electrophoresis in birds. Am J Vet Res. 2005;66(3):375-378.
(17.) Johnston MS, Rosenthal KL, Shofer FS. Assessment of a point-of-care biochemical analyzer and comparison with a commercial laboratory for the measurement of total protein and albumin concentrations in psittacines. Am J Vet Res. 2007;68(12): 1348-1353.
Carolyn Cray, PhD, Ed King, PhD, Marilyn Rodriguez, BS, MT, Lilli S. Decker, DVM, and Kristopher L. Arheart, EdD
From the Division of Comparative Pathology, Department of Pathology (Cray, Rodriguez), Department of Epidemiology and Public Health (Arheart), University of Miami Miller School of Medicine, PO Box 016960 (R-46), Miami, FL 33101, USA; and the ALX Laboratory, Animal Medical Center, 510 E 62nd St, New York, NY 10065, USA (King, Decker).
Table 1. Summary of EPH fraction values by electrophoresis system. Mean (95% CI) EPH value Beckman, g/dL Helena, g/dL Prealbumin 0.49 (0.40-0.57) 0.59 (0.53-0.65) Albumin 1.97 (1.79-2.16) 1.68 (1.56-1.81) [[alpha].sub.1] Globulins 0.12 (0.09-0.14) 0.14 (0.13-0.16) [[alpha].sub.2] Globulins 0.14 (0.11-0.16) 0.17 (0.16-0.19) [beta] Globulins 0.59 (0.54-0.65) 0.76 (0.67-0.85) [gamma] Globulins 0.32 (0.29-0.34) 0.29 (0.26-0.32) A : G ratio 2.14 (2.02-2.27) 1.73 (1.62-1.85) Mean (95% CI) EPH value Sebia, g/dL Prealbumin 0.13 (0.11-0.15) Albumin 2.05 (1.89-2.21) [[alpha].sub.1] Globulins 0.61 (0.54-0.67) [[alpha].sub.2] Globulins 0.09 (0.08-0.11) [beta] Globulins 0.42 (0.39-0.46) [gamma] Globulins 0.32 (0.30-0.35) A : G ratio 1.52 (1.41-1.62) Abbreviations: EPH indicates protein electrophoresis; CI, confidence interval; A : G, albumin to globulin. Table 2. Total coefficient of variation for protein fractions using plasma from healthy psittacine birds and the Beckman, Helena, and Sebia electrophoresis systems. Beckman, Helena, Sebia, Fraction % (n = 18) (a) % (n = 20) % (n = 24) Prealbumin 4.3 9.1 7.5 Albumin 3.2 2.6 2.2 [[alpha].sub.1] Globulins 9.3 9.5 1.8 [[alpha].sub.2] Globulins 14.8 12.0 7.5 [beta] Globulins 4.2 3.5 4.0 [gamma] Globulins 9.1 8.3 4.0 (a) Results from Cray et al. (7) Table 3. Summary of results of statistical analyses comparing the Beckman method with the Helena method using plasma samples from African grey parrots (n = 40). Passing-Bablok regression analysis Fraction r value y-intercept 95% CI Prealbumin 0.72 (a) -0.21 -0.51 to -0.04 Albumin 0.93 (a) -0.15 -0.45 to 0.04 [[alpha].sub.1] Globulins 0.79 (a) -0.05 -0.15 to 0.0 [[alpha].sub.2] Globulins 0.75 (a) -0.08 -0.15 to -0.05 [beta] Globulins 0.25 (b) -0.20 -0.33 to -0.04 [gamma] Globulins 0.81 (a) -0.03 -0.09 to 0.04 Passing-Bablok regression analysis Propor- Constant tional Fraction error? Slope 95% CI error? Prealbumin Yes 1.29 1.00 to 1.80 No Albumin No 1.22 1.11 to 1.43 Yes [[alpha].sub.1] Globulins No 1.14 0.75 to 1.91 No [[alpha].sub.2] Globulins Yes 1.20 1.00 to 1.67 No [beta] Globulins Yes 1.16 0.92 to 1.36 No [gamma] Globulins No 1.14 0.90 to 1.42 No Abbreviation: CI indicates confidence interval. (a) Pearson correlation coefficient was significant, P < .05. (b) Pearson correlation coefficient was not significant, P > .05. Table 4. Summary of results of statistical analyses comparing the Beckman method with the Sebia method using plasma samples from African grey parrots (n = 40). Passing-Bablok regression analysis Fraction r value y-intercept 95% CI Prealbumin 0.09 (a) -2.29 Not defined (c) Albumin 0.89 (b) -0.30 -0.80 to 0.04 [[alpha].sub.1] Globulins 0.47 (b) 0.01 -0.03 to 0.04 [[alpha].sub.2] Globulins 0.65 (b) 0.00 -0.07 to 0.04 [beta] Globulins 0.46 (b) -0.35 -0.71 to -0.15 [gamma] Globulins 0.84 (b) -0.12 -0.25 to -0.05 Passing-Bablok regression analysis Constant Fraction error? Slope 95% CI Prealbumin Yes 24.4 Not defined (c) Albumin No 1.09 0.89 to 1.37 [[alpha].sub.1] Globulins No 0.14 0.09 to 0.21 [[alpha].sub.2] Globulins No 1.54 1.00 to 2.36 [beta] Globulins Yes 2.30 1.80 to 3.14 [gamma] Globulins Yes 1.33 1.13 to 1.73 Passing-Bablok regression analysis Proportional Fraction error? Prealbumin Yes Albumin No [[alpha].sub.1] Globulins Yes [[alpha].sub.2] Globulins No [beta] Globulins Yes [gamma] Globulins Yes Abbreviation: CI indicates confidence interval. (a) Pearson correlation coefficient was not significant, P > .05. (b) Pearson correlation coefficient was significant, P < .05. (c) Not defined by statistical software because of marked difference between methods. Table 5. Summary of Bland-Altman plot data comparing the Beckman method with the Helena method or the Sebia method. Beckman versus Helena % Differences Mean within difference, imprecision Fraction g/dL limits Prealbumin -0.10 72.5 Albumin 0.29 32.5 [[alpha].sub.1] Globulins -0.02 25.0 [[alpha].sub.2] Globulins -0.04 60.0 [beta] Globulins -0.16 75.0 [gamma] Globulins 0.03 80.0 Beckman versus Sebia % Differences Mean within difference, imprecision Fraction g/dL limits Prealbumin 0.35 5.0 Albumin -0.07 20.0 [[alpha].sub.1] Globulins -0.49 0 [[alpha].sub.2] Globulins 0.04 57.5 [beta] Globulins 0.17 15.0 [gamma] Globulins -0.01 72.5
|Printer friendly Cite/link Email Feedback|
|Author:||Cray, Carolyn; King, Ed; Rodriguez, Marilyn; Decker, Lilli S.; Arheart, Kristopher L.|
|Publication:||Journal of Avian Medicine and Surgery|
|Date:||Jun 1, 2011|
|Previous Article:||Serum biochemical values of adult ostriches (Struthio camelus) anesthetized with xylazine, ketamine, and isoflurane.|
|Next Article:||Surveys of Avian practitioners and pet owners regarding common behavior problems in psittacine birds.|