Introduction
Blood (plasma and serum) TP concentration in veterinary medicine is measured routinely by means of external laboratory and in-clinic biochemistry analyzers, which use the biuret assay (spectrophotometry) and detect peptide bonds.1 This assay is considered to have high diagnostic accuracy for this purpose.1 Alternatively, plasma or serum TP concentration can be estimated as TS concentration with the less expensive, more accessible, and faster refractometry method, by which the protein molecules in plasma or serum increase the refractive index of the fluid in proportion to the concentration of these molecules.2 Differences can be expected in TP measurements performed with these 2 methods, regardless of the type of sample (plasma or serum) tested.1,2 In addition, refractometer readings for TS concentration in plasma samples can exceed those for paired serum samples, owing in part to the absence of fibrinogen in serum.1
Much debate exists in the veterinary literature concerning the potential interference of sample hemolysis, lipemia, colloids administration, or high sample concentrations of bilirubin, BUN, glucose, chloride, or sodium on measurements of TS concentration by refractometer.1,2,3,4,5 It has been suggested that, in the absence of these factors, the amount of nonprotein solids present in plasma is fairly constant (1.5 g/dL); therefore, the plasma TP concentration can be estimated by subtracting 1.5 from the refractometer TS reading.3 Because of potential disagreement with TP concentration, the possibility exists that TS concentration may not be used interchangeably,6 and method and species-specific data are necessary for accurate interpretation of test results.7
The purpose of the study reported here was to compare paired plasma TP and TS concentrations in pet rabbits (Oryctolagus cuniculus) and ferrets (Mustela putorius furo) as measured by biuret assay and refractometry, respectively, to determine whether the 2 methods could be used interchangeably and whether the refractometer could provide an accurate estimation of plasma TP concentration in these 2 species. The study hypothesis was that there would be no difference between measured values for TP and TS concentration, regardless of patient species and signalment, PCV, plasma analyte concentrations, or plasma color.
Materials and Methods
Animals
Medical records of pet rabbits and ferrets presented for various health evaluations to the Exotic and Zoo Animal Medicine Service of the Veterinary Health Center, Kansas State University from 2008 through 2018 were reviewed to determine eligibility for the study. For inclusion, animals were required to have had both a CBC and plasma biochemical analyses performed at the same time and under similar testing conditions during this period. The target minimum sample size was 31 of each species, which was comparable to sample sizes reported for similar research in other species.6
Data collection
Data were collected for each patient regarding sex, neuter status, and breed (for rabbits only). Results of plasma sample analysis and CBC were also obtained. Multiple results from the same animal were excluded if they were obtained during the same medical event; in that situation, only the first set was included. However, multiple results from the same patient were included if the tests did not occur during the same medical event.
Sample collection and analysis
Blood samples had been collected as part of the diagnostic work-up for each patient. To facilitate collection, ferrets were anesthetized with 2% isofluranea in oxygen delivered via face mask. Immediately following induction, a 1-mL syringe with a 25-gauge, 16-mm needle was used to obtain the blood sample from the cranial vena cava. Rabbits were manually restrained without anesthesia, and the blood sample was obtained from the lateral saphenous vein with a similar syringe and needle.
Immediately after collection, blood samples were transferred to tubesb containing lithium heparin. Samples for plasma TS analysis were processed routinely with the microhematocrit method, which involved centrifugation for 4 minutes at 13,500 × g. Samples for plasma biochemical analyses were centrifuged for 5 minutes at 906 × g. After plasma was harvested, the TS concentration was measured with a temperature-compensated veterinary refractometer.c Plasma TP concentration was measured by biuret assay with an automated biochemistry analyzer.d Measurements were performed concurrently at the Kansas State Veterinary Diagnostic Laboratory within 2 hours after sample collection.
The color of each plasma sample was assessed visually and with the biochemistry analyzer, which reported a number (index value) for each sample. Samples were then mixed with saline (0.9% NaCl) solution and measured with the biochemistry analyzer at various wavelengths according to the index for hemolysis, lipemia, and icterus.
Quality control for the refractometer consisted of weekly assessment of the refractive index by use of distilled water. The biochemistry analyzerd was maintained and calibrated in accordance with the manufacturer's instructions and internal laboratory standard operating procedures. Instrument performance (internal quality control) was monitored daily by use of commercial quality control materiale and 1–2s or 1–3s control rules,8 depending on the analyte measured. Every 3 months, performance of the biochemistry analyzer was evaluated as part of an external quality assurance program.f All laboratory analyses were performed by trained medical laboratory technologists.
The combined imprecision of the 2 analytic methods was determined by repeated analysis (6 times) of 3 residual plasma samples/species/method. Additionally, imprecision at low concentrations was assessed by dilution of one of the residual plasma samples from each species with 50% saline (NaCl) solution and analysis of the diluted sample 6 times. The species-specific CV for each method was determined by analyzing 4 rabbit or ferret samples 6 times each and then calculating the mean CV for the results.
Statistical analysis
For study purposes, analytic bias was defined as the biuret assay (reference method) value for plasma TP concentration minus the refractometer value for plasma TS concentration. Because some animals contributed > 1 sample to the analyses and the effects of several biochemical variables and PCV on bias were of interest, bias was further evaluated by use of a linear mixed-model approach for each species. Signalment and biochemical variables as well as mean values for the 2 analytic methods ([refractometer value + biuret value]/2; for assessment of proportional bias) were added to the model as fixed effects, and patient was added as a random effect. The 95% LOA were calculated as bias ± 1.96 × σ, where σ is the within-subject (ie, residual) SD obtained from the linear mixed model.
The combined imprecision of the 2 methods9 was calculated as the square root of (CVrefractometer2 + CVbiuret2). A TEa value of 10% was used for measurements of plasma TP concentration.10
A modified Bland-Altman plot11 was generated, including the data points, patient mean values (because some animals had several data points), mean bias, 95% LOA, and analytic acceptance limits based on the combined inherent imprecision of the 2 analytic methods. Pearson correlation coefficients were also computed to assess correlation between methods. The Shapiro-Wilk test was used to assess normality of data distribution.
Methods were considered analytically equivalent (analytic agreement) if the LOA were within the analytic acceptance limits. Techniques were considered clinically comparable (clinical agreement) if the LOA were within the TEa.
Values of P < 0.05 were considered significant. All statistical tests were performed and graphs generated12 with freely available statistical software.g,h
Results
Animals
Overall, 146 ferrets and 121 rabbits were included in the study. Of the 146 ferrets, 49% were female (58 spayed and 13 sexually intact) and 51% were male (60 neutered and 15 sexually intact). Of the 121 rabbits, 40% were female (31 spayed and 18 sexually intact), 58% were male (48 neutered and 22 sexually intact), and 2 were of unknown sex. Rabbit breeds as reported by the owners included lop eared (n = 33), domestic (27), Dutch Dwarf (16), Mini Rex (12), Lionhead (9), Mini Lop (8), Netherland Dwarf (7), Rex (7), and Satin (2). The 146 ferrets contributed 253 blood samples to the study, and the 121 rabbits contributed 146 blood samples.
Samples
Hemolysis was noted in 27 of the 146 (18.5%) rabbit plasma samples; the remaining 119 (81.5%) samples were unremarkable in color. Hemolysis was noted in 48 (19.0%) of the 253 ferret plasma samples, lipemia in 18 (7.1%), mixed hemolysis-lipemia in 8 (3.2%), icterus in 4 (1.6%), and mixed icterus-lipemia in 1 (0.4%); the remaining 174 (68.8%) samples were unremarkable.
Not accounting for repeated measurements, respective mean ± SD values for plasma TP and TS concentrations were 6.4 ± 0.8 mg/dL and 6.6 ± 0.8 mg/dL for rabbits and 6.3 ± 1.2 mg/dL and 6.4 ± 1.1 mg/dL for ferrets.
Comparisons between measurement methods
A significant (P < 0.001) constant negative bias (with plasma TS values overestimating plasma TP values on average) was observed for both ferrets (mean bias, −0.32 mg/dL; 95% LOA, −1.26 to 0.62 mg/dL) and rabbits (mean bias, −0.20 mg/dL; 95% LOA, −0.98 to 0.58 mg/dL). Cholesterol concentration, glucose concentration, BUN concentration, hemolysis, and lipemia had significant effects on the magnitude of the bias for ferrets (Table 1). For rabbits, hemolysis and lipemia could not be evaluated because no samples had these properties; however, BUN concentration had a significant impact on measurement bias. No proportional bias was identified for either species. No influence on measurement bias was identified for sex, neuter status, breed (rabbits only), and other biochemical analytes. The Pearson correlation between TP and TS values for ferrets and rabbits was 0.87 and 0.88, respectively.
Parameter estimates (mean ± SEM) derived from linear mixed models showing the effect of various plasma analytes on the mean bias* between values for plasma TS concentration as measured by refractometry and plasma TP concentration as measured by biuret assay (reference method) in ferrets and rabbits.
Ferrets (n = 253 samples) | Rabbits (n = 146 samples) | |||
---|---|---|---|---|
Analyte | Mean ± SEM | P value | Mean ± SEM | P value |
Cholesterol (per 100 mg/dL) | −0.29 ± 0.07 | < 0.001 | −0.12 ± 0.09 | 0.19 |
Glucose (per 100 mg/dL) | −0.10 ± 0.04 | 0.023 | −0.09 ± 0.05 | 0.07 |
BUN (per 10 mg/dL) | −0.03 ± 0.01 | 0.001 | −0.04 ± 0.01 | 0.001 |
Hemolysis (per 1-point increase in index value) | 0.26 ± 0.09 | 0.006 | 0.09 ± 0.10 | 0.39 |
Lipemia (per 1-point increase in index value) | −0.23 ± 0.09 | 0.007 | NA | NA |
NA = Not applicable.
Bias was calculated as the biuret value minus the refractometer value. Overall mean bias for ferret samples was −0.32 mg/dL and for rabbit samples was −0.20 mg/dL.
The 95% LOA (equivalent to TEobs) for both species were wider than the TEa and acceptance limits, indicating that the 2 analytic methods were in neither analytic nor clinical agreement (Figures 1 and 2). A TEa of 10% based on their respective mean TP values was 0.63 mg/dL (ferrets) and 0.64 mg/dL (rabbits), which we considered the clinically accepted cutoff and lower than the TEobs.10 Overall, plasma TS concentration as measured by refractometry failed to adequately estimate the plasma TP concentration as measured by biuret assay for both ferrets and rabbits, even after accounting for confounding biochemical effects and correcting for bias.
Method imprecision
The CV (imprecision) of refractometer measurements for ferrets and rabbits was 0.6% and 1.1%, respectively, and the CV of biuret assay measurements was 4.6% and 1.4%, respectively. The CV for both methods combined (combined imprecision) was 4.6% for ferrets and 1.7% for rabbits.
Discussion
Refractometer measurements of plasma TS concentration are often used to estimate plasma TP concentration in clinical veterinary medicine.2 However, findings of the present study suggested that these 2 methods cannot be used interchangeably for rabbits and ferrets because the measured differences between the 2 methods exceeded values for TEa.10 A significant constant negative bias, whereby the refractometer values overestimated the biuret values on average, was observed for both species, with no proportional bias identified for either species. Many studies, including those involving ruminants,6 dogs and cats,13,14 horses,15 and chickens and turkeys,16 have shown variable degrees of correlation between plasma TP and TS concentrations that required a correction formula for estimation of TP concentration from the TS readings, if the bias was observed to be constant or linear. In the present study, although the Pearson correlation between plasma TP and TS values was fairly high at 0.87 and 0.88, no correction formula could be applied because the LOA were wider than the analytic acceptance limits.
Despite a few reports that indicate plasma TP measurements are higher than plasma TS measurements in cattle,6 goats,6 pigs,17 dogs,18 cats,18 and horses,18 most reports7,13,14,15,16,19,20,21,22,23,24 in veterinary medicine indicate that plasma TS concentration typically overestimates plasma TP concentration. This overestimation has many possible explanations. The refractive index depends on the soluble particles in the tested solution, and their concentration can interfere with TS readings.25 Blood urea nitrogen concentration increased the measurement bias between methods used in the present study for both ferrets and rabbits; this effect was similarly observed for dogs and cats.20 Plasma glucose and cholesterol concentrations also increased measurement bias for ferrets. These findings suggested that caution should be exercised when attempting to estimate TP concentrations from TS concentrations in the presence of high cholesterol, glucose, and BUN concentrations.
The effects of lipemia, hemolysis, and icterus on measurements of plasma TS concentration have been evaluated in other species. In a study22 involving dogs, these characteristics had no effect on plasma TS readings. Another study13 involving dogs and cats revealed higher variability in plasma TS readings when lipemia and hemolysis were present, and this variability was greater in dogs than in cats. A third study20 involving cattle, sheep, horses, dogs, and cats showed that evidence of lipemia, but not icterus, in plasma samples was associated with higher TS readings and that hemolysis can artifactually interfere with the clarity of the refractometer scale reading. None of the rabbit plasma samples in the present study had evidence of lipemia or icterus, so the effect of these variables on plasma TP and TS measurements could not be determined. For ferrets, however, hemolysis and lipemia had an effect on plasma TS measurements, suggesting the effect of hemolysis on refractometer readings might be species specific.
In the study reported here, fresh plasma derived from heparinized blood samples was evaluated, rather than serum from clotted blood or a comparison of plasma TS concentration with serum TP concentration. This choice was intentional because the concentration of TS in plasma is slightly higher than that in serum (approx 5%) owing to the aforementioned absence of fibrinogen in serum.3 The type of anticoagulant used for plasma preparation can also affect the results and should also be considered when comparing methods for TP quantitation.21,25,26 Fresh plasma samples were used in the present study because storage can also have an effect on measurements of TP and TS concentrations in animals.22,27
Limitations of the study reported here included its retrospective nature and reliance on medical record data. All laboratory testing had been performed by trained technicians at an accredited laboratory, but some interoperator variability that was not accounted for in the statistical analyses could have existed over the 11-year study period. In addition, the findings may be specific to the instruments used for measurement or estimation of plasma TP concentration.7,16,21,28
In conclusion, refractometer measurements of plasma TS concentration overestimated values for plasma TP concentration as measured by biuret assay for the ferrets and rabbits of the present study. The limits of agreement (TEobs) exceeded the 10% TEa, indicating that the 2 methods for TP measurement were not in clinical agreement and cannot be used interchangeably in ferrets and rabbits.
Acknowledgments
The authors thank Dr. Diana Schwartz for advice on this project.
AbbreviatioNS
CV | Coefficient of variation |
LOA | Limits of agreement |
TEa | Total allowable analytic error |
TEobs | Total observed analytic error |
TP | Total protein |
TS | Total solids |
Footnotes
IsoFlo, Abbott Laboratories, North Chicago, Ill.
BD Microtainer tube, Becton, Dickinson and Co, Franklin Lakes, NJ.
Leica 10436, Kernco Instruments Co, El Paso, Tex.
Cobas c501, Roche Diagnostics, Indianapolis, Ind.
PreciNorm and Precipath, Roche Products Ltd, Roche Diagnostics, Indianapolis, Ind.
Veterinary Laboratory Association Quality Assurance Program, Atlantic Veterinary College, University of Prince Edward Island, PE, Canada.
R: a language and environment for statistical computing, version 3.5.2, R Foundation for Statistical Computing, Vienna, Austria. Available at: www.r-project.org. Accessed Jun 22, 2019.
nlme: Linear and nonlinear mixed effects models, R package, version 3.1–140. Available at cran.r-project.org/package=nlme. Accessed Jun 22, 2019.
References
- 1. ↑
Allison RW. Laboratory evaluation of plasma and serum proteins. In: Thrall MA, ed. Veterinary hematology and clinical chemistry. 2nd ed. Oxford: Blackwell Publishing, 2012;460–475.
- 2. ↑
George JW. The usefulness and limitations of hand-held refractometers in veterinary laboratory medicine: an historical and technical review. Vet Clin Pathol 2001;30:201–210.
- 3. ↑
Melillo A. Applications of serum protein electrophoresis in exotic pet medicine. Vet Clin North Am Exot Anim Pract 2013;16:211–225.
- 4. ↑
Yam E, Hosgood G, Rossi G, et al. Synthetic colloid fluids (6% hydroxyethyl starch 130/0.4 and 4% succinylated gelatin) interfere with total plasma protein measurements in vitro. Vet Clin Pathol 2018;47:575–581.
- 5. ↑
Yam E, Boyd CJ, Hosgood G, et al. Hydroxyethyl starch 130/0.4 (6%) and succinylated gelatine (4%) interfere with refractometry in dogs with haemorrhagic shock. Vet Anaesth Analg 2019;46:579–586.
- 6. ↑
Katsoulos PD, Athanasiou LV, Karatzia MA, et al. Comparison of biuret and refractometry methods for the serum total proteins measurement in ruminants. Vet Clin Pathol 2017;46:620–624.
- 7. ↑
Cray C, Rodriguez M, Arheart KL. Use of refractometry for determination of psittacine plasma protein concentration. Vet Clin Pathol 2008;37:438–442.
- 8. ↑
Westgard QC. “Westgard Rules” and multirules. Available at: www.westgard.com/mltirule.htm. Accessed Aug 29, 2020.
- 9. ↑
Jensen AL, Kjelgaard-Hansen M. Method comparison in the clinical laboratory. Vet Clin Pathol 2006;35:276–286.
- 10. ↑
Harr KE, Flatland B, Nabity M, et al. ASVCP guidelines: allowable total error guidelines for biochemistry. Vet Clin Pathol 2013;42:424–436.
- 11. ↑
Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Biochim Clin 1987;11:399–404.
- 12. ↑
Wickham H. ggplot2: elegant graphics for data analysis. 2nd ed. New York: Springer-Verlag, 2016.
- 13. ↑
Briend-Marchal A, Médaille C, Braun JP. Comparison of total protein measurement by biuret method and refractometry in canine and feline plasma. Rev Med Vet 2005;156:615–619.
- 14. ↑
Hayes GM, Mathews K, Floras A, et al. Refractometric total plasma protein measurement as a cage-side indicator of hypoalbuminemia and hypoproteinemia in hospitalized dogs. J Vet Emerg Crit Care (San Antonio) 2011;21:356–362.
- 15. ↑
Carlson GP, Harrold DR. Relationship of protein concentration and water content of equine serum and plasma samples. Vet Clin Pathol 1977;6:18–20.
- 16. ↑
Andreasen CB, Latimer KS, Kircher IM, et al. Determination of chicken and turkey plasma and serum protein concentrations by refractometry and the biuret method. Avian Dis 1989;33:93–96.
- 17. ↑
Green SA, Jenkins SJ, Clark PA. A comparison of chemical and electrophoretic methods of serum protein determinations in clinically normal domestic animals of various ages. Cornell Vet 1982;72:416–426.
- 18. ↑
Lackhoff A, Walden A. Comparative study of total plasma protein concentration in the dog, cat and horse by the biuret and refractometry methods. Berl Munch Tierarztl Wochenschr 1984;97:8–10.
- 19. ↑
Legendre KP, Leissinger M, Le Donne V, et al. The effect of urea on refractometric total protein measurement in dogs and cats with azotemia. Vet Clin Pathol 2017;46:138–142.
- 20. ↑
Sutton RH. The refractometric determination of the total protein concentration in some animal plasmas. N Z Vet J 1976;24:141–148.
- 21. ↑
Tamborini A, Papakonstantinou S, Brown A, et al. Comparison of manual and laboratory PCV and total protein using EDTA and lithium heparin canine samples. J Small Anim Pract 2014;55:258–264.
- 22. ↑
Arfuso F, Giannetto C, Rizzo M, et al. Comparison of refractometric and biuretic methods for the assay of total protein in horse serum and plasma under various storage conditions. J Equine Vet Sci 2018;61:58–64.
- 23. ↑
Lumeij JT, De Bruijne JJ. Evaluation of the refractometric method for the determination of total protein in avian plasma or serum. Avian Pathol 1985;14:441–444.
- 24. ↑
Lumeij JT, Maclean B. Total protein determination in pigeon plasma and serum: comparison of refractometric methods with the biuret method. J Avian Med Surg 1996;10:150–152.
- 25. ↑
Cannon DC, Olitzky I, Inkoen JA. Proteins. In: Henry RJ, Cannon DC, Winkelman JW, eds. Clinical chemistry principles and technics. 2nd ed. Hagerstown, Md: Harper & Row, 1974;407–421.
- 26. ↑
Dubin S, Hunt P. Effect of anticoagulants and glucose on refractometric estimation of protein in canine and rabbit plasma. Lab Anim Sci 1978;28:541–544.
- 27. ↑
McSherry BJ, Al-Baker J. Comparison of total serum protein determined by t/s meter and biuret technique. Vet Clin Pathol 1976;5:4–12.
- 28. ↑
Dawnay AB, Hirst AD, Perry DE, et al. A critical assessment of current analytical methods for the routine assay of serum total protein and recommendations for their improvement. Ann Clin Biochem 1991;28:556–567.