Reference intervals for electrophoretograms obtained by capillary electrophoresis of dialyzed urine from healthy dogs

Abstract

Electrophoresis of urine to evaluate protein fractions in dogs with proteinuria to differentiate glomerular from tubular damage has increased in recent years; however, capillary electrophoresis (CE) of urine has not been reported in a study of > 40 healthy animals, to our knowledge. We aimed to establish reference intervals (RIs) for the urine protein fractions obtained by CE of urine from healthy dogs. We obtained urine samples from 123 clinically healthy dogs of both sexes between December 2016 and April 2019; urine was frozen until CE was performed. The electrophoretic patterns obtained were divided into 5 protein fractions, and RIs were established in percentages and absolute values using nonparametric methods. RIs were obtained for the fractions (F) as follows: 5.5 to 56.2% for F1, 3.2 to 16.5% for F2, 3.5 to 16.2% for F3, 17.8 to 69.8% for F4, and 5.1 to 23.9% for F5. These RIs obtained by CE might be useful clinically as a basis for comparison with pathologic samples. Age was a statistically significant factor for F2 (p = 0.01) and F3 (p = 0.02), and sex was a statistically significant factor for F1 (p = 0.03).

Introduction

Serum protein analysis by capillary electrophoresis (CE) is a well-established laboratory method used for the diagnosis and follow-up of infectious, inflammatory, immune-mediated, and neoplastic conditions in human and veterinary medicine, given that it is a simple, adaptable, and rapid technique that does not require a large amount of sample.^9,18,25 The CE process is based on the separation of charged molecules based on their electrophoretic mobility in an alkaline buffer at a specific pH. The separation occurs according to the buffer pH and electroosmotic flow, yielding various electrophoretic fractions.^26,32 In plasma or serum, the electrophoretogram in mammals is commonly divided into the following fractions from low to high molecular weight (MW) and charge: albumin, alpha-1-, alpha-2-, beta-1-, beta-2-, and gamma globulins.^5,30

Analysis of proteins in urine by CE has proven to be a suitable method in human medicine to detect the presence of characteristic electrophoretic patterns in renal and extrarenal disorders.^17,24,31 Although only quantitative proteinuria can be assessed by the urine protein:creatinine ratio (UPC), electrophoretic techniques can be used as a semiquantitative method to assess the loss of proteins through urine because different patterns can be identified. ¹⁴

Pathologic renal proteinuria is the presence of protein in urine attributable to structural or functional lesions within the kidneys. The origin of this pathologic proteinuria can be glomerular, tubular, or interstitial. ²¹ Normally, the small amount of protein present in filtrate passes through the glomerular capillary walls and is reabsorbed by the proximal tubules. Proteins larger than albumin (69 kDa) rarely pass through glomeruli. ¹³ A few studies have examined the most abundant proteins in the urine of healthy animals. In 2010, a study detected C-reactive protein, immunoglobulin G, thromboxane B2, and retinol-binding protein in the urine of healthy dogs. ²³

Different methods can be used to evaluate proteinuria in dogs, including colorimetric strips, the UPC ratio, and urine albumin and microalbumin concentrations. ¹¹ The most common method used today in veterinary medicine to measure proteinuria is the UPC ratio given that it provides a quantitative result. Physiologic proteinuria in canine species is considered when the UPC ratio is < 0.2, borderline proteinuria is considered when the ratio is 0.2 to 0.5, and significant proteinuria is considered when the ratio is > 0.5, provided that it is accompanied by an inactive urinary sediment. ²¹

A few studies in veterinary medicine have correlated different urine electrophoretogram patterns with pathologic conditions using sodium dodecyl sulfate agarose–polyacrylamide gel electrophoresis (SDS-AGE; SDS-PAGE) and high-resolution gel electrophoresis (HRE).^{2,10,14,20,33,35} To our knowledge, no data are available concerning CE electrophoretic patterns in normal or pathologic urine from dogs. We chose CE because of its advantages in serum, including the reduced separation time (2 min vs. 20 min in a gel), reduction of waste harmful to humans and the environment, and ability to obtain a curve without the need for additional staining as is needed for gel electrophoresis.^8,9,12,22 Moreover, urinary CE in dogs is new and may be of interest to veterinary laboratorians.

We aimed to establish reference intervals (RIs) for the various protein fractions obtained from the urine of healthy dogs by CE for comparison with pathologic conditions.

Materials and methods

Twenty-three veterinary centers, including veterinary clinics and reference veterinary hospitals, collaborated between December 2016 and April 2019. Our study was approved by the research and ethics committee of the Universidad Católica de Valencia San Vicente Mártir (Valencia, Spain; UCV 2017-2018-33). Urine samples were obtained from 123 clinically healthy dogs by free catch, which could be done easily by the owners and did not require specific training. A potential disadvantage of free catch, instead of cystocentesis, is the hypothetical presence of proteins from the reproductive system, given that we also included intact animals in our study. The inclusion criteria for the study were biological data for the patient, such as age (juvenile = 0–1 y old; adult = > 1–8 y old; senior = > 8 y old), sex (female, male), reproductive status (intact, castrated, or spayed), breed, and clinical parameters such as clinical history (no signs of illness in the 2 wk preceding or following sample collection because longer follow-up was not possible in some cases), and a routine physical examination.

The study population was located in Spain and included outdoor and indoor dogs. Preventive care information was not available for all patients; therefore, it was not included as a study factor. Routine hematology, serum biochemistry, urinalysis, and serology to detect antibody titers for Leishmania infantum, Ehrlichia canis, and Rickettsia conorii were performed. Data for urine protein concentration were only recorded in 110 dogs. We excluded from our study patients with an abnormal history or physical examination and/or concurrent diseases and those that were apparently healthy but subsequently developed clinicopathologic alterations.

The dogs were divided into 4 subgroups by age range, sex, and reproductive status, differentiating males from females (Table 1). Several breeds were included randomly in our study: mixed-breed (56), Boxer (8), Dachshund (6), Labrador Retriever (6), Yorkshire Terrier (6), Golden Retriever (4), Spanish Hound (4), Maltese Bichon (3), Ratonero Valenciano (3), Staffordshire Bull Terrier (3), Doberman Pinscher (2), German Shepherd (2), Greyhound (2), and Siberian Husky (2). Breeds represented by a single dog each included Basset Hound, Beagle, Belgian Shepherd, Border Collie, Chihuahua, French Bulldog, German Shorthair Pointer, Pomeranian, and Pug.

Table 1.

Age and sex of the 123 healthy dogs studied to evaluate dialyzed urine fractions by capillary electrophoresis.

	M	F	Intact (M/F)	Castrated/spayed	Unknown (M/F)
Puppy (1 mo–1 y)	15	18	12/17	3/0	0/1
Adult (1–8 y)	19	34	6/10	13/18	0/6
Senior (> 8 y)	20	17	4/2	12/12	4/3
Total	54	69	22/29	28/30	4/10

F = female; M = male.

Blood samples were collected from either the cephalic or jugular vein after 12 h of fasting. Samples of 0.5 mL of whole blood in a 0.5-mL EDTA tube (Aquisel) and a minimum of 1.5 mL of blood in a 2-mL tiger-top tube for serum collection (Aquisel) were obtained, centrifuged (Centrifuge 2650; Nahita) at 1,340g for 10 min, refrigerated in 1.5-mL microcentrifuge tubes (Lambda) at 4°C, and analyzed within the following 24 h. Owners were asked to collect a minimum of 8 mL of the first urine in the morning by free catch of a mid-stream sample in a 150-mL sterile plastic specimen cup (Deltalab) supplied by the hospital. Urine was stored at 4°C and analyzed within the following 12 h.

All EDTA blood, sera, and urine samples were processed by a reference laboratory (Cedivet, Valencia, Spain). The analysis included a complete blood count (CBC; Celltac Alpha VET MEK-6550; Nihon) with blood smear evaluation. Biochemical analytes measured included creatinine, urea, alanine aminotransferase, and total serum proteins (CS 300 analyzer; Dirui). Serum electrophoretograms were run by CE (Minicap instrument; Sebia). Sera of all animals were also tested for antibodies to L. infantum, E. canis, and R. conorii by an immunofluorescent antibody test (Axio Scope HBO 50 microscope; Zeiss) because these infectious disease agents are highly prevalent in the Mediterranean area.

Urine samples were centrifuged (Centrifuge 2650; Nahita) at 804g for 5 min, and the supernatants were divided into 1-mL aliquots, minimum of 4 mL, and stored at −20°C prior to dialysis in a 1.5-mL plastic microcentrifuge tube (Lambda). An aliquot of freshly collected urine (maximum of 24 h after collection) was kept to evaluate the following variables: (1) microscopic fresh and stained sediment at low- and high-power fields (Binocular microscope DM500; Leica); (2) urine total proteins and creatinine (automated chemistry analyzer; Dirui), to allow calculation of the UPC (urine creatinine dilution was 1:49 in all samples; creatinine was measured by a creatinine enzymatic method, and proteins in urine were measured with pyrogallol red reagent); (3) urine culture in a specific chromogenic medium (CHROMagar orientation medium; Becton Dickinson); (4) specific gravity (refractometer; Optika Ponteranica); and (5) pH, glucose, ketone, bilirubin, and hemoglobin or myoglobin levels (LabStrip U11 Plus; 77 Elektronika).

Urine was dialyzed before CE (Sebia) to eliminate compounds that could interfere with the wavelength used for reading and cause artifact peaks, and to avoid salts that could have deleterious effects on the capillary tube. The dialysis procedure followed the manufacturer’s protocol. Urine supernatant (4 mL) from each dog was thawed at room temperature and centrifuged at 1,609 × g for 10 min. The resulting supernatant was transferred to a 4-mL dialysate column (Vivaspin Turbo 4 10,000 MWCO; Sartorius). Dialysate columns containing urine were centrifuged at 1,878 × g for 25 min; the urine that emerged at the filtrate container was discarded. Washing solution was prepared adding 50% ultrapure distilled water and 50% dialysis buffer (Sebia) in a sterile container. The concentrator was refilled to 4-mL volume with this solution and centrifuged at 1,878g for 20 min. The 200 μL obtained was transferred to a 1.5-mL microcentrifuge tube and subjected to CE (Minicap; Sebia).

Twenty-six samples could be run at once, and 100 μL was sufficient to perform the analysis. Silica capillaries were filled with protein separation buffer (Sebia). Then the samples were aspirated into the anodic end of the capillary. High-voltage was applied to run the sample along the capillary. When samples reached the cathodic end, detection and quantification of the protein fractions were performed at a wavelength of 200 nm. An electrophoretic curve was obtained for analysis.

As quality control (QC) material, we included frozen aliquoted sera from a healthy dog, diluted in running buffer at 1:49 and migrated prior to any run and in each batch. The manufacturer reported a sensitivity of 20.0 mg/L, and a maximum coefficient of variation (CV) of 4% from the area comprised of the baseline of the chart to the peak apex of the electrophoretogram, validated for monoclonal proteins, which may vary depending on the polyclonal background.

Internal verification experiments were also performed; within-run and between-run experiments were performed using urine from a healthy dog. This urine was stored at 4°C during the entire experiment and dialyzed and concentrated for every run following the manufacturer’s protocol. Three migrations per sample and day (repeatability) were made for 5 consecutive days (reproducibility), with the objective of calculating the CV for each of the fractions of the urinary proteinogram. As QC material, a 1:49 diluted serum from the same dog was run simultaneously. The limit of detection (LOD) was determined with the urine sample of the same healthy dog. Briefly, 1 in 2, 1 in 4, 1 in 8, and 1 in 16 dilutions of the dialyzed urine were made in running buffer and were run. The LOD was selected as the last dilution with an electrophoretic pattern equal to the undiluted urine.

The urine electrophoretic pattern obtained for each dog was divided into protein fractions 1 to 5 (F1–F5). These fractions were determined by superimposing the normal canine diluted serum used as QC material over the electrophoretic samples (instead of human serum as suggested by the manufacturer in the human urine protocol). Protein fractions were verified and, if necessary, corrected by visual inspection of the electrophoretogram. All samples were analyzed by the same person (P. Navarro).

Outliers were detected (Reference value advisor macro v.2.0; Microsoft). ⁷ Histograms of the reference values of each urine electrophoretogram fraction were assessed to identify potential outliers (Figs. 1, 2). Tukey interquartile fences and the Dixon outlier range statistic were used to identify true and suspected outliers. True outliers were deleted, whereas suspected outliers were retained following the American Society for Veterinary Clinical Pathology (ASVCP) guidelines. ⁴

Figure 1.

Histograms used to assess distribution in percentages of fractions 1 to 5 by capillary electrophoresis in dialyzed urine from healthy dogs. Two outlier values were eliminated from F2. X axis values are percentages. The dashed line boxes represent 90% CIs of upper and lower limits of the RI of each fraction.

Figure 2.

Distribution in absolute values (mg/L) of fractions 1 to 5 by capillary electrophoresis in dialyzed urine from healthy dogs. Two outlier values were eliminated from F2, F3, and F5, and one outlier value was eliminated from F4. X axis values are in mg/L. The dashed line boxes represent 90% CIs of upper and lower limits of the RI of each fraction.

RIs were obtained (Reference value advisor v.2.0; Microsoft) through the nonparametric method given that the population was large enough. ⁷ Information regarding data distribution was not necessary because nonparametric methods were used. Moreover, 95% RIs in percentages and absolute values were calculated with 90% CIs for reference limits according to the ASVCP 2012 guidelines. Bootstrapping was used when the population was < 120.^4,6 Partitioning analysis by age and reproductive status was not possible given the low number of dogs in each subgroup (< 40). ⁴ Partitioning by sex was rejected according to the z-score. ¹⁹

Statistical analysis to compare the different subgroups was performed (R v.3.4.3; https://www.r-project.org/). The Anderson–Darling test was used for each subgroup to test the hypothesis of normality. Outliers were detected and eliminated if they were considered aberrant observations, although the emphasis was to retain them rather than delete them. The ANOVA test was used to establish whether age, sex, or reproductive status, or differentiating males from females, influenced the proteinogram fractions.

Results

In the within-run experiment, the CV results were 3.38%, 3.84%, 7.25%, 4.43%, and 7.31% for F1, F2, F3, F4, and F5, respectively. In the between-run experiment, the CV results were 4.78%, 5.17%, 10.0%, 6.09%, and 9.66% for F1, F2, F3, F4, and F5, respectively. The sensitivity obtained in the LOD experiment was 2.1 mg/L.

Electrophoretogram fractions F2 and F3 were decreased significantly in puppies compared to the adult and senior groups (Table 2). F1 was increased significantly in females compared to males (Table 3). No significant differences were found between intact and castrated males in the 5 urinary electrophoretogram fractions; no significant differences were found between intact or spayed females in any fraction. From percentage data, 2 outliers were deleted from F2. Regarding absolute data, 2, 2, 1, and 2 outliers were removed from F2, F3, F4, and F5, respectively. The set of the remaining dogs was used to obtain the RIs of each fraction of the urine curve (percentage and absolute). RIs were obtained for the different fractions as follows: 5.5 to 56.1% for F1, 3.2 to 16.5% for F2, 3.5 to 16.2% for F3, 17.8 to 69.8% for F4, and 5.1 to 23.9% for F5 (Table 4). RIs in absolute values (mg/L) were obtained for the fractions, with the following values: 2.49 to 138 mg/L for F1, 0.87 to 36.3 mg/L for F2, 1.23 to 33.5 mg/L for F3, 6.68 to 182 mg/L for F4, and 2.32 to51.5 mg/L for F5 (Table 5). A urine electrophoretogram from a healthy dog with 5 different fractions is shown in Figure 3. The RI obtained for the UPC (n = 123) was 0.0 to 0.3. The RI for total proteins in urine (n = 109) was 14 to 408 mg/L.

Table 2.

Effects of age on canine dialyzed urine electrophoretogram analytes by capillary electrophoresis evaluated with a one-way ANOVA (Tukey post hoc test with 95% CI).

Analyte	N	Factor	Mean	SD	p-value*
F2	121	Puppy	6.87	2.37	0.01
		Adult	8.55	2.37
		Senior	8.85	3.62
F3	122	Puppy	7.23	2.35	0.02
		Adult	8.85	2.79
		Senior	8.53	2.84

n = 123 dogs (53 adults, 33 puppies, 37 seniors, except when outliers were excluded); normality testing was assessed using the Anderson–Darling test with adjusted Holm test. ANOVA = analysis of variance; CI = confidence interval; SD = standard deviation.

^*Significance p ≤ 0.05.

Table 3.

Effects of sex on canine dialyzed urine electrophoretogram analytes by capillary electrophoresis evaluated with a one-way ANOVA (Tukey post hoc test with 95% CI).

Analyte	N	Factor	Mean	SD	p-value*
F1	123	Male	25.6	14.3	0.03
		Female	31.1	13.6

n = 123 dogs (69 females, 54 males, except when outliers were excluded); normality testing was assessed using the Anderson–Darling test with adjusted Holm test.

ANOVA = analysis of variance; CI = confidence interval; SD = standard deviation.

^*Significance p ≤ 0.05.

Table 4.

Reference intervals in percentages for dialyzed urinary capillary electrophoresis fractions using Minicap (Sebia) in dogs.

Analyte	Unit	n	Mean	SD	Median	IQR	Min.	Max.	RI	LRL (90% CI)	URL (90% CI)	Distribution	Method
F1	%	123	28.7	14.0	27.0	19.6–39.2	2.5	58.3	5.54–56.2	2.50–6.70	52.2–58.3	Gaussian	NP
F2	%	121	8.19	2.91	8.10	6.2–9.9	1.8	17.2	3.21–16.5	1.80–3.80	12.6–17.2	Gaussian	NP
F3	%	123	8.41	2.87	8.10	6.6–9.6	3.3	17.4	3.51–16.2	3.30–4.10	13.4–17.4	Non-Gaussian	NP
F4	%	123	43.0	13.0	42.5	33.2–51.7	9.2	75.7	17.8–69.8	9.20–22.7	65.0–75.7	Gaussian	NP
F5	%	123	11.2	4.48	10.5	8.05–13.1	4.5	26.4	5.11–23.9	4.50–5.70	20.0–26.4	Non-Gaussian	NP

n = 123 dogs, except when outliers were excluded. CI = confidence interval; IQR = interquartile range; LRL = lower reference limit; NP = nonparametric; RI = reference interval; SD = standard deviation; URL = upper reference limit.

Table 5.

Reference intervals in absolute values for dialyzed canine urinary capillary electrophoresis fractions using Minicap (Sebia) in dogs.

Analyte	Unit	n	Mean	SD	Median	IQR	Min.	Max.	RI	LRL (90% CI)	URL (90% CI)	Distribution	Method
F1	mg/L	110	4.92	3.63	4.38	2.06–6.82	1.11	141	2.49–138	11.1–42.0	131–142	Non-Gaussian	NP
F2	mg/L	108	1.43	0.87	1.29	0.80–1.95	0.64	41.0	0.87–36.3	0.64–2.21	31.0–41.0	Non-Gaussian	NP
F3	mg/L	108	1.40	0.81	1.25	0.84–1.93	0.46	37.3	1.23–33.5	0.46–2.29	29.0–37.3	Non-Gaussian	NP
F4	mg/L	109	7.45	4.30	7.35	4.35–10.2	2.30	216	6.68–182	2.30–16.7	154–216	Non-Gaussian	NP
F5	mg/L	108	1.88	1.18	1.64	1.02–2.39	0.49	57.1	2.32–51.5	0.49–4.69	45.2–57.1	Non-Gaussian	NP

n = 110 dogs, except when outliers were excluded. CI = confidence interval; IQR = interquartile range; LRL = lower reference limit; NP = nonparametric; RI = reference interval; SD = standard deviation; URL = upper reference limit.

Figure 3.

Example electrophoretogram by capillary electrophoresis in dialyzed urine with 5 fractions differentiated; results given in percentages and absolute values. Urine collected by free catch from a healthy 4-y-old, spayed female Golden Retriever. Urine protein:creatinine ratio = 0.10. The values of all fractions fall within the RIs found in our study.

Discussion

The UPC ratios of all samples included in our study were < 0.5, and all sediments were inactive. This UPC range is consistent with non-proteinuric IRIS (International Renal Interest Society) staging of proteinuria. We observed no statistically significant relationship in the urine electrophoretic pattern between entire or neutered male or female animals. These results could suggest that, although prostatic antigen arginine esterase is known to be found in urine of intact males, it may not be excreted constantly or in a significant amount, or it might be below the instrument’s LOD. Nevertheless, studies with a larger population would be necessary to verify this suggestion.^15,20,28

Statistical differences found between age and sex in some fractions were significant. F2 and F3 were found to be decreased in puppies compared to adult and senior dogs. In serum, these 2 fractions are related to globulins alpha₁ and alpha₂ ³⁰ ; therefore, this decrease could be related to immunity development during growth or to the administration of some recent preventive therapy for which information was not available to us. ²⁷ F1 was found to be significantly higher in bitches compared to males. It may be that obesity and sex played a role in the urinary metabolome. Additional studies with a larger and homogeneous population would be necessary to confirm these hypotheses.^1,3

To our knowledge, RIs for CE in the urine of healthy dogs over a wide age range, with this number of animals, and with a high confidence level have not been reported previously. RIs of urine CE in healthy dogs could be used to assess pathologic urinary samples, as used in human medicine to evaluate patients with kidney disease and paraproteinemia. Future lines of study include comparison with urine from dogs with renal disease in which differences between fractions would be expected, such as an increase in the excretion of F1 in dogs with chronic kidney disease. Anti-IgG has been detected in the urine of dogs with leishmaniasis and chronic kidney disease ³⁴ ; therefore, it would be reasonable to think that F5 would be increased if renal damage occurred in infectious diseases. Comparing our results with the data obtained in other studies on the identification of urine proteins is difficult because the techniques used in those studies are different, and the numbers of animals in the samples are very diverse. Qualitative studies have been conducted in animals with kidney disease at different stages of severity, primarily in search of biomarkers.^{2,10,14,15,20,22} A study was conducted in dogs with kidney disease in an attempt to differentiate patterns of renal tubular and glomerular disease by SDS-PAGE. ²⁹ Another study ³⁵ compared the patterns obtained with renal histologic lesions using SDS-PAGE in dogs. Using immunoblotting, another study ³³ compared the bands obtained by SDS-PAGE in healthy dogs to those observed in animals with renal proteinuria; the most abundant protein in healthy animals was albumin (MW 65 kDa). Other proteins were also detected in lesser amounts, such as α1-microglobulin (MW 27 kDa) and transferrin (MW 76 kDa). All of these studies detected mixed glomerular and tubular patterns, showing high (> 65 kDa) and low (< 65 kDa) MW bands, and tubular patterns showing only low MW bands.

A limitation in our study is the inability to compare CE results with the results of other techniques using gel electrophoresis, such as SDS-PAGE. Proteins in CE migrate based on their charge and size; proteins migrate only based on their size in SDS techniques. Establishment of the different fractions has also been challenging because we found only one reference related to the establishment of the urinary fractions, and it was focused on paraproteins. ¹⁶ A follow-up period of > 2 wk to ensure the lack of renal disease was not possible in the majority of our cases. No data were recorded on preventive care, weight, and body condition score in our cases, factors that could potentially alter protein excretion in urine from healthy animals.