Detection of Canine Microalbuminuria Using Semiquantitative Test Strips Designed for Use with Human Urine

Microalbuminuria refers to increased urine albumin concentrations that are lower than those theoretically measurable by standard urine dipsticks.¹ Microalbuminuria in humans is defined as urine albumin excretion between 20 and 200 μg/min or a concentration of 20–200 μg/mL at normal diuresis.¹ In humans, microalbuminuria is a predictor of late onset clinical proteinuria and eventual end-stage renal disease.^1,2 Microalbuminuria occurs with increased frequency in humans with insulin-dependent and non-insulin-dependent diabetes mellitus, hypertension, neoplasia, cardiovascular disease, and acute systemic insults such as surgery, pancreatitis, or myocardial infarction.³ Detection of microalbuminuria may allow early intervention, thus preventing or slowing progression to overt glomerular disease.¹

Measurement of urinary albumin excretion rate requires timed urine collection, usually over a 24-hour period, and quantification by radioimmunoassay, ELISA, immunoturbidometry, or nephelometry.⁴ To allow instantaneous determination of microalbuminuria, urine albumin concentration also can be calculated from a single voided sample; however, this method assumes normal urine production and output.⁵ A calculated urine albumin:creatinine ratio is highly correlated with urine albumin concentration measured by timed urine collection and corrects for variability in diuresis and urine concentration.^5,6 Because of the ease of a single urine collection versus timed collection, human patients are screened for microalbuminuria using one of a number of commercially available point-of-care semiquantitative test strips, which determine albumin concentration or albumin:creatinine ratio, with further quantitative testing via timed urine collection as indicated.^4–6

A canine albumin-capture ELISA has been used to determine that microalbuminuria occurs in both overtly healthy dogs and dogs with a variety of clinical illnesses or conditions.^7,8 Therefore, the validation of dipsticks for point-of-care testing of canine urine samples may provide practitioners with a valuable diagnostic tool for detection of early glomerular disease. The purpose of this study was to determine the concordance of canine urine albumin concentrations measured by a canine albumin-capture ELISA and by a commercial test strip designed for use with human urine.

Materials and Methods

Urine samples were prospectively collected from dogs at the North Carolina State University Veterinary Teaching Hospital. Dogs with a variety of clinical conditions were randomly selected from the hospital patient population between February 2000 and August 2000. However, to ensure a wide range in albumin concentration in urine samples, additional effort was made to collect urine from dogs known or thought to have overt proteinuria. A majority of dogs sampled were those that were presented for outpatient evaluation, ie, dogs whose clinical condition did not necessitate hospitalization.

Method of urine collection was chosen by the attending clinician; collection method was not used as a basis for exclusion but was recorded for later analysis. Urine sediment examination and/or urine culture also were done at clinician discretion; all dogs had urine dipstick analysis (Multistix, Bayer Corporation, Elkhart, Ind, USA). Hematuria was defined as >5 RBC/high power field (×400) on urine sediment examination or >1+ blood on dipstick urinalysis. Urinary tract infection was diagnosed if quantitative urine culture had >1000 colony-forming units (cfu)/mL in samples obtained by cystocentesis.

A commercially available urine dipstick was used in this study. The Clinitek Microalbumin reagent strip (Bayer) is composed of 2 reaction pads that allow a visual estimate of albumin and creatinine concentrations using color comparison charts. The dipstick can also be evaluated with a desktop reflectance photometer such as the Clinitek 50 or Clinitek 100 (Bayer), which reports albumin and creatinine concentrations and calculates the albumin:creatinine ratio. The albumin:creatinine ratio is reported as <30 μg/mg, 30–300 μg/mg, or >300 μg/mg, which is interpreted by the manufacturer as negligible urine albumin concentration, possible microalbuminuria, or overt albuminuria, respectively. All urine samples in this study were processed immediately following collection in strict accordance with the manufacturer's recommendations. Dipstick results were evaluated with the Clinitek 50 urine chemistry analyzer.

Aliquots of urine were frozen at −80°C within 15 minutes of collection. Urine albumin concentration was subsequently measured using a canine albumin-capture ELISA.⁷ Microtiter plates (MaxiSorp, Nalge Nunc International, Rochester, NY, USA) were coated overnight at 4°C with rabbit anti-canine albumin IgG. After coating, microtiter plates were washed with phosphate-buffered saline (PBS) containing 0.05% polyoxyethylene (20) sorbitan monolaurate (Tween 20, JT Baker, Phil-lipsburg, NJ, USA) (PBS-T), blocked (StabilCoat, Sur-modics Corp, Eden Prarie, Minn, USA) for 1 hour at ambient temperature, and then washed again with PBS-T. Four washes were performed at each step throughout the ELISA preparation. Samples (100 μL) were added to the microtiter wells and incubated for 2 hours at ambient temperature. After washing with PBS-T, 100 μL of biotinylated goat anti-canine albumin IgG (Bethyl Laboratories Inc, Montgomery, Tex, USA; 125 ng/mL) was added to each well and incubated for 30 minutes at ambient temperature. After washing with PBS-T, 100 μL of peroxidase-labeled streptavidin (500 ng/mL) was added to each well and incubated for 30 minutes at ambient temperature. After a final PBS-T wash, 100 μL of tetramethyl benzidine substrate (TMB Peroxidase Substrate System, Kirkegaard and Perry Laboratories, Gaithersburg, Md, USA) was added to each well and incubated for 30 minutes at ambient temperature. The reaction was stopped with 1 M phosphoric acid, and optical density (OD) at 450 nm was measured for each well.

Negative controls consisted of nonpooled urine samples from 86 privately-owned clinically normal dogs.⁷ Positive controls were prepared by adding 100 μg canine albumin per milliliter of negative control urine. Canine albumin standards were prepared by serially diluting canine albumin in PBS containing 0.1% casein (PBS-C); the range of the standard curve was 1.9–30.0 ng/mL. Canine urine samples and positive and negative controls were serially diluted 2-fold from 1:500 to 1:3200 in PBS-C so that reactions would occur within the linear range of the standard curve.

Background negative control well values were subtracted from all sample well values. An OD versus canine albumin concentration standard curve was generated using the canine albumin standard OD values. Albumin concentrations for the unknown urine samples were determined by extrapolation from the standard curve and correction for the initial sample dilution. ELISA albumin concentration results were normalized to a urine specific gravity (SpGr) of 1.010 to allow comparison between samples. Normalization was performed using the following formula: normalized albumin concentration = measured albumin concentration × (0.01/[SpGr-1.0]).

Analytical imprecision of the ELISA was verified using the prepared standards. Intra-assay coefficient of variation was 2.0%-17.0% (3 replicates). Interassay variation was 4.5%-21.7% (assay performed 21 days over 11 weeks). The ELISA was able to detect all canine albumin added to a sample (100 μg/mL sample; recovery mean ± SD = 109.99 ± 18.94 μg/mL).

Using the canine albumin-capture ELISA, urine albumin concentrations <10 μg/mL were classified as negative, and microalbuminuria was defined as 10–300 μg/mL.⁷ Urine albumin concentrations >300 μg/mL (ie, theoretically detectable by standard urine dipsticks) were classified as overt albuminuria. Samples were not analyzed for concordance based on patient underlying disease.

Results

Urine samples were collected from 67 dogs. Sample collection was by cystocentesis (27 dogs), catheterization (7 dogs), or midstream void (33 dogs). Urine sediment examinations were performed in 58 dogs. Urine dipstick analyses were performed on all urine samples; urine cultures were performed on samples from 19 dogs.

Urine albumin concentration ranged from 0.1 to >500 μg/mL as determined by the albumin-capture ELISA. Of the 67 samples tested, 25 had negligible urine albumin concentration (<10 μg/mL), 27 had microalbu-minuria-range albuminuria (10–300 μg/mL), and 15 had overt albuminuria (>300 μg/mL). Urine samples from 7 dogs did not indicate proteinuria by dipstick analysis; 3 of these dogs had microalbuminuria and 1 had overt albuminuria as determined by ELISA.

The Clinitek strips correctly reported urine albumin concentration in 42 of 67 (63%) urine samples tested. Urine albumin concentration was correctly determined in 19 (76%) dogs with negligible urine albumin, 13 (48%) dogs with microalbuminuria, and 10 (67%) dogs with overt albuminuria (Figure 1). Of the 14 dogs with microalbuminuria that were misclassified by the Clinitek strips, 7 (50%) had negligible albuminuria and 7 (50%) had overt albuminuria. Sensitivity and specificity of the Clinitek strips for correct identification of microalbuminuria in urine samples were 48% and 75%, respectively (Table 1). Sensitivity and specificity of the Clinitek test strips for correct identification of increased urine albumin concentration (>10 μg/mL) were 83% and 76%, respectively.

Results obtained by the Clinitek strip and ELISA were in concordance in 19 of 27 (70%) samples obtained by cystocentesis, 2 of 7 (29%) samples obtained by catheterization, and 21 of 33 (64%) voided samples. Results from the Clinitek strips and ELISA were in concordance in 0 of 2 dogs with confirmed bacterial urinary tract infection. One other dog had very low growth (<100 cfu/mL) of a group G Streptococcus sp; results from this dog were concordant between Clinitek strips and ELISA. Three dogs had bacteria on urine sediment examination (1+ to 2+) but did not have urine cultures performed. The results for only 1 of these 3 dogs were concordant between the Clinitek strips and ELISA.

Clinitek test strip and ELISA results were in concordance in 9 of 21 (43%) dogs with hematuria. Exclusion of urine samples from dogs with confirmed urinary tract infection or hematuria minimally improved Clinitek strip sensitivity (59%) and specificity (83%) for correct identification of microalbuminuria (Table 2). Sensitivity and specificity for identification of dogs with increased urinary albumin levels but without urinary tract infection or hematuria were 83% and 77%, respectively.

Discussion

Point-of-care diagnostic tests ideally identify a majority of animals with a given abnormality or disease (ie, have high sensitivity) and thus allow clinicians to make informed decisions as to which animals require further diagnostic testing. Therefore, a point-of-care microalbuminuria test strip should reliably identify animals with microalbuminuria-range albuminuria, alerting the clinician to possible early glomerular disease. The results in this study suggest that Clinitek test strips lack sufficient concordance with results obtained by quantitative canine albumin-capture ELISA and lack sufficient sensitivity to be reliable point-of-care screening tests for urine albumin levels in dogs.

In addition to the Clinitek test strips, urine samples were initially tested with a second commercial product, the Micral test strip (Roche Diagnostics Corporation, Indianapolis, Ind, USA; data not shown). These test strips consist of a series of reagent pads that produce an antibody-enzyme complex that reacts with a single color-substrate pad. Final pad color is compared with colors on a reference chart. Results are reported as negative or within a series of ranges from 20 to >100 mg/L (μg/mL). Of the 13 urine samples evaluated, 12 (92%) were negative for urine albumin, and concordance with ELISA results occurred in only 4 (31%) samples. Further testing with this product was discontinued.

Both the Clinitek and Micral test strips rely upon antibody-impregnated pads that bind to human albumin, producing a visual reaction. Canine and human albumin are highly homologous^9,10; however, the test strips we evaluated may have failed because of differences between these 2 proteins. Alternatively, there may be a factor in canine urine that alters the test strip antibody-antigen complex formation or color change reaction. We used the calculated albumin:creatinine ratio to determine the concentration of urine albumin predicted by the Clinitek test strip. Although the molecular structure of creatinine does not differ between dogs and humans, some factor in canine urine may alter the expected human creatinine test strip pad reactions.

Although catheterization clearly resulted in the lowest concordance between the Clinitek test strip and ELISA results, no collection method resulted in satisfactory concordance. In human medicine, urine is most commonly collected by the patient as a midstream voided sample. However, in our study, voided samples showed less concordance than did samples collected by cystocentesis. Urinary tract infection and hematuria also appeared to negatively impact concordance of results obtained by the Clinitek test strip and ELISA. However, exclusion of samples from dogs with these conditions did not adequately improve the Clinitek test sensitivity or specificity to make it reliable for use.

The Clinitek test strip evaluated in this study was not sufficiently accurate for use as a screening tool for microalbuminuria in dogs. Exclusion of patients with conditions that may have interfered with test strip accuracy did not sufficiently improve sensitivity or specificity. Given the recognition of microalbuminuria as a clinical entity, a reliable point-of-care screening test for low concentrations of urine albumin is still needed.