Comparison of urine dipstick, sulfosalicylic acid, urine protein-to-creatinine ratio and a feline-specific immunoassay for detection of albuminuria in cats with chronic kidney disease

Abstract

The performance of the urine dipstick, sulfosalicylic acid (SSA), and urine protein-to-creatinine (UPC) tests for the detection of albuminuria was assessed in cats with chronic kidney disease (CKD). Two hundred and thirty-nine urine samples from 37 cats with CKD were used. Test results were dichotomized as either positive or negative, compared with those for the feline-specific rapid urine albumin immunoassay and test performance variables calculated for each test. A positive urine dipstick (≥trace) and positive SSA (≥5 mg/dl), positive SSA alone or ≥2+ urine dipstick alone were indicative of albuminuria. In these cases, protein quantification would be warranted if proteinuria/albuminuria is persistent. In the case of a negative urine dipstick result the addition of the SSA added little diagnostic value. Of the tests investigated, the single best test for the detection of albuminuria was the UP/C (≥0.2) in which either a negative or positive test result provided useful information.

Introduction

Chronic kidney disease (CKD) is common in cats, especially aged ones. Proteinuria may be associated with CKD in cats and when present proteinuria is a risk factor for disease progression and death.^1–4 Attenuation of proteinuria in cats with CKD may be associated with improved clinical signs and azotemia, slowed disease progression and/or improved survival.^3,5,6 Detection and longitudinal monitoring of proteinuria in cats with CKD is an important diagnostic tool.

Normally, feline urine contains less than 1 mg/dl of albumin. ⁷ The term proteinuria is reserved to indicate excess protein in the urine. ⁸ Multiple proteins can be found; however, albumin is the most abundant protein in urine in healthy cats, as well as in cats with CKD.^9,10 Microalbuminuria (MA) is a low concentration of urine albumin. For the purposes of this study, MA has been defined as a urine albumin concentration of 1–30 mg/dl in urine that has been diluted to a specific gravity of 1.010.^7,8,11 Proteinuria/albuminuria may be caused by pre- and post-renal disease conditions in addition to intrinsic renal disease; therefore, localization of the source of proteinuria is an important diagnostic consideration.^7,8,11

The traditional urine dipstick colorimetric test is the most commonly used screening test for proteinuria in cats, but false-positive reactions for protein limit its utility^9,10 and therefore the sulfosalicylic acid (SSA) turbidity test is used by many laboratories to confirm positive dipstick reactions for proteinuria.^10,12 When proteinuria detected by the dipstick and/or SSA screening tests is suspected to be of renal origin, the urine protein-to-creatinine ratio (UPC) or a species-specific albumin immunoassay [enzyme-linked immunosorbent assay (ELISA)] is used to both confirm and quantify proteinuria.^8,12 Microalbuminuria may be detected by either a quantitative or semiquantitative feline-specific immunoassay (ERD health screen feline urine test, Heska Corporation). Compared with a feline-specific quantitative urine albumin immunoassay, both the urine dipstick and SSA tests performed poorly and appeared to be of minimal diagnostic value for the detection of albuminuria in healthy cats because of an unacceptable number of false-positive results. ¹⁰ In addition, in healthy cats, the UPC was highly specific, but lacked sensitivity. ¹⁰ For that reason, a species-specific ELISA has been recommended for the detection of albuminuria in cats. ¹⁰

Both the ERD and quantitative immunoassay have been shown to be more sensitive and specific for the detection of systemic disease compared with other methods, such as the urine dipstick and UPC. ¹³ Both assays performed very similarly, with the sensitivity being 43.1% and 36.9% for the ERD and quantitative immunoassay, respectively. Positive test results were found in 19% and 25% of the 441 cats tested with the ERD and quantitative immunoassay, respectively. In a recently published, multicenter study, the quantitative immunoassay was used as the ‘gold-standard’ for the detection of albuminuria in the healthy cat. ¹⁰ Even more recently, the ERD has been used to describe the prevalence of MA in cats with diabetes mellitus. ¹⁴

The purpose of the study reported here is to assess the diagnostic performance of the urine dipstick, SSA and UPC compared with the ERD for detection of albuminuria in cats with CKD. We hypothesized that owing to higher concentrations and/or greater frequency of albuminuria expected in cats with CKD, the UPC, SSA and urine dipstick would be more accurate for the detection of albuminuria compared with that reported previously for healthy cats. ¹⁰

Materials and methods

Animals

The sample population consisted of cats (n = 37) with naturally-occurring CKD (International Renal Interest Society stages 2 and 3) being evaluated and treated as part of a clinical trial at the College of Veterinary Medicine at Kansas State University between February 2008 and July 2011. A diagnosis of CKD was based on a combination of the following clinical findings: chronic history (≥2 months) of weight loss and or polydipsia/polyuria; non-regenerative anemia; persistent azotemia (initial serum creatinine concentration ≥1.6 but≤3.6 mg/dl) in the face of dilute urine [urine specific gravity (USG) <1.025]; abdominal radiograph and ultrasound evaluations; indirect blood pressure determination (Doppler Ultrasonic Flow Detector, Parks Medical Electronics); urine culture; UPC; and total thyroid hormone (TT₄). Cats with USG >1.025 were required to remain azotemic in the euhydrated state and have structural renal changes (small irregular kidney(s) and/or loss of cortico-medullary definition) evident on abdominal ultrasound. Exclusion criteria included uncontrolled hyperthyroidism, positive bacterial urine culture, evidence of obstructive uropathy, fractious nature or evidence of other concurrent disease (eg, diabetes mellitus, neoplasia). Informed and signed owner consent was obtained before entry into the study. The study protocol in its entirety was approved by the Animal Care and Use Committee at Kansas State University. Once enrolled in the trial, cats were evaluated initially at 1 month and then subsequently every 3 months. All owners were instructed to withhold food for 12 h prior to each examination. It was recommended to make water available at all times.

Urine assays

Urine samples were collected via cystocentesis. Complete urinalyses were performed at the Clinical Pathology Laboratory at Kansas State University. Urine samples were analyzed the day of collection; if not performed immediately the urine sample was refrigerated for no more than 2 h. Urine samples were excluded from further albuminuria/proteinuria analysis when urine sediment examination revealed any of the following abnormalities: >200 red blood cells/high power field (hpf), >5 white blood cells/hpf, spermaturia or bacteriuria.

Urinalysis was performed with a commercially-available urine dipstick (Multistix 10 SG, Siemens Healthcare Diagnostics). The color change on the urine dipstick was interpreted by a commercially-available analyzer with normal and abnormal controls run daily (Clinitek Status Analyzer, Siemens Healthcare Diagnostics). Scores for protein concentration on the dipsticks were recorded as negative, trace, 1+, 2+, 3+ or 4+. The SSA tests were performed by mixing equal volumes of urine supernatant and SSA (5%) in a glass tube and grading the resulting turbidity as negative or ≥5 mg/dl compared with a set of gel standards. The UPC was calculated by dividing the urine total protein concentration (determined by the benzethonium chloride reaction by use of an automated chemistry analyzer (COBAS - C501, Hitachi 911; Roche Diagnostics) by the urine creatinine concentration (determined via the buffered kinetic Jaffe reaction by use of an automated chemistry analyzer; COBAS - C501, Hitachi 911).

The coefficient of variation (CV) for one chemistry analyzer (COBAS - C501) was 4.1% and 4% with a mean creatinine concentration of 74.3 mg/dl and 136.1 mg/dl, respectively. The CV of the same chemistry analyzer was 6% and 2% for a mean protein concentration of 15 mg/dl and 46 mg/dl, respectively. The CV for the second chemistry analyzer (Hitachi 911) was 1.9% and 4% with a mean creatinine concentration of 80.9 mg/dl and 12 mg/dl respectively. The CV for the same chemistry analyzer was 4% and 1% with a mean protein concentration of 12 mg/dl and 45 mg/dl, respectively. The lower and upper limit of detection for both chemistry analyzers was 4–200 mg/dl and 4.2–622 mg/dl for protein and creatinine, respectively. For microprotein concentrations >200 mg/dl the urine sample was not diluted, but a smaller sample (2 μl instead of 6 μl) was analyzed. For creatinine measurement urine samples were prediluted by the chemistry analyzer with 6 μl of urine being added to 144 μl of 0.9% NaCl. For creatinine concentrations >622 mg/dl, 6 μl of urine was diluted with 180 μl of 0.9% NaCl. There is a high correlation concerning protein and creatinine concentrations as measured on the two chemistry analyzers, with a correlation coefficient (R) of 0.9990 and 0.9972, respectively (data not shown).

Urine albumin concentration for each urine sample was measured with a feline-specific semiquantitative rapid immunoassay (ERD) within 30 mins of collection. Urine samples were not refrigerated. USG was determined by means of a refractometer (HSK-VET, Heska Corporation). To account for the varying urine concentrations, samples were normalized to a specific gravity of 1.010 using distilled water as a diluent. Scoring for the urine albumin concentration was recorded as negative, low positive, medium positive, high positive or very high positive based on the relative intensities of two colored bands on the test device, as described in the package insert.

Statistical analysis

Analyses were performed using commercial software (STATA 11; Stata Corporation). Albumin concentrations as determined by ERD were used as the criterion reference standard for this study; all data were dichotomized as either positive or negative. All ERD reactions except for negative ones were considered positive (low, medium, high and very high positive) test results. The urine dipstick result was considered positive for all results with ≥trace reaction. An additional analysis was conducted with cut-offs, categorizing dipstick positives ≥2+ and ≥3+. SSA test results were categorized as positive for all results ≥5 mg/dl and the UPC was categorized as positive for all results ≥0.2 and ≥0.4. Sensitivity, specificity, negative predictive value (NPV), positive predictive value (PPV), likelihood ratio positive (LR+) and likelihood ratio negative (LR−) were calculated for each test. A combination of urine dipstick and SSA test performance was assessed in both series (interpreted as positive only when both test results were positive) and parallel (interpreted as positive when either test result was positive). A receiver operator curve (ROC) was generated for the urine dipstick (≥trace), SSA (≥5 mg/dl), and UPC (≥0.2 and ≥0.4). The area-under-curve (AUC) for each ROC was compared with a one-way ANOVA and Bonferonni post-hoc adjustment for multiple comparisons (six total tests resulting in a critical P value of 0.0083).

Test performance

A useful test is one in which the result significantly increases (positive test result) or decreases (negative test result) the probability that a disease/physiologic state exists in the tested cat. Here, the disease state is albuminuria, as defined as a positive test result on ERD. The larger the LR+ the more likely a positive sample is from a truly albuminuric animal. The smaller the LR− the less likely a negative sample is from a truly albuminuric animal. Diagnostic recommendations in this study are based on the calculated LR+ and LR− for each test with consideration given to the other calculated test performance variables when appropriate.

Results

Two hundred and thirty-nine urine samples from 37 cats fulfilled the inclusion criteria and were included in the study (Table 1). The median number of urine samples from each cat was 6.0 with a range of 1–15.

Table 1.

Clinicopathologic findings and number of urine samples from 37 cats diagnosed with chronic kidney disease and included in this study

Cat	Number of urine samples	Mean HCT	Mean CREA	Mean UPC	Mean USG
1	7	36	2.5	0.09	1.013
2	14	26	2.4	0.26	1.013
3	4	30	2.8	0.50	1.011
4	12	38	2.8	1.20	1.011
5	9	37	2.6	0.13	1.020
6	6	28	3.1	0.13	1.013
7	12	34	3.5	0.12	1.014
8	5	38	1.6	0.10	1.015
9	11	32	3.3	0.35	1.019
10	15	33	1.9	0.13	1.021
11	15	34	2.0	0.47	1.010
12	12	31	2.1	0.63	1.018
13*	6	40	1.6	0.28	1.034
14*	12	42	1.7	0.49	1.034
15*	8	37	1.9	0.13	1.032
16	5	17	5.5	0.26	1.006
17	7	29	2.9	0.10	1.016
18	7	24	2.9	2.50	1.015
19	4	38	2.3	0.15	1.012
20	9	41	2.2	0.10	1.014
21	9	34	2.8	0.12	1.013
22	10	21	2.9	0.15	1.012
23	13	38	2.5	0.09	1.017
24	9	30	2.7	0.10	1.013
25	3	25	2.7	0.36	1.011
26	2	29	3.3	0.20	1.020
27	2	44	1.6	0.25	1.021
28	2	40	1.4	3.25	1.017
29	1	40	2.9	0.30	1.026
30	1	28	2.1	0.30	1.016
31	1	32	2.7	0.50	1.009
32	1	28	2.5	0.50	1.016
33	1	28	3.7	0.20	1.013
34	1	34	2.8	0.20	1.018
35	1	31	3.0	0.20	1.013
36	1	N/A	3.2	0.10	1.016
37	1	38	2.1	0.10	1.013
Total mean	6.45	33	2.5	0.37	1.017

^*Cats with USG >1.025 were required to have persistent azotemia in euhydrated state and structural renal changes consistent with CKD on abdominal ultrasound

HCT = hematocrit (%); CREA = serum creatinine (mg/dl); UPC = urine protein-to-creatinine ratio; USG = urine specific gravity as determined by refractometry; N/A = not available

Results for the assessment of the urine dipstick and SSA tests when classified as positive for all test results of ≥trace or ≥5 mg/dl, respectively, are summarized in Table 2. Performance for the UPC is summarized in Table 3. Performance of the urine dipstick when the definition of a positive test was increased from ≥trace to ≥2+ and ≥3+ are summarized in Table 4.

Table 2.

Performance of the urine dipstick (≥trace) and/or sulfosalicyclic acid (SSA) test (≥5 mg/dl) to detect albuminuria as identified by the semiquantitative feline-specific immunoassay (ERD) (≥low positive) in the same urine samples

Variable	Positive urine dipstick (≥trace)	Positive SSA test (≥5 mg/dl)	Positive urine dipstick and positive SSA test	Positive urine dipstick or positive SSA test
Sensitivity (%)	81.2	62.9	59.5	84.5
Specificity (%)	68.0	95.9	95.9	68.0
PPV (%)	70.9	93.6	93.2	71.5
NPV (%)	79.0	73.1	71.3	82.2
LR+	2.54	15.4	14.5	2.64
LR−	0.28	0.39	0.42	0.23

PPV = positive predictive value; NPV = negative predictive value; LR+ = likelihood ratio positive; LR− = likelihood ratio negative

Table 3.

Performance of the urine protein-to-creatinine ratio (UPC; ≥0.2 and ≥0.4) to detect albuminuria as identified by the semiquantitative feline-specific immunoassay (ERD) (≥low positive) in the same urine samples

Variable	UPC ≥0.2	UPC ≥0.4
Sensitivity (%)	84.6	45.3
Specificity (%)	81.8	100.0
PPV (%)	81.8	100.0
NPV (%)	84.6	65.4
LR+	4.65	.
LR−	0.19	0.55

PPV = positive predictive value; NPV = negative predictive value; LR+ = likelihood ratio positive; LR− = likelihood ratio negative

Table 4.

Performance of the urine dipstick (≥trace, ≥2+ and ≥3+) to detect albuminuria as identified by the semiquantitative feline-specific immunoassay (ERD) (≥trace) in the same urine samples.

Variable	Urine dipstick ≥trace	Urine dipstick ≥2+	Urine dipstick ≥3+
Sensitivity (%)	81.2	42.7	12.0
Specificity (%)	68.0	97.5	100.0
PPV (%)	70.9	94.3	100.0
NPV (%)	79.0	64.0	54.2
LR+	2.54	17.4	.
LR−	0.28	0.59	0.88

PPV = positive predictive value; NPV = negative predictive value; LR+ = likelihood ratio positive; LR− = likelihood ratio negative

The ROCs for the urine dipstick, SSA and UPC test results compared with results for the ERD are illustrated in Figure 1, and the AUC for the ROC analysis for each test is summarized in Table 5. The AUC for UPC (≥0.2) was significantly different than the AUC for the UPC (≥0.4) (0.8315 vs 0.7241; P = 0.0002). The AUC for the UPC (≥0.2) was not significantly different than that for SSA (0.8315 vs 0.7940; P = 0.1849) or urine dipstick (≥trace) (0.8315 vs 0.7481; P = 0.0121). The AUC for SSA was significantly larger than the AUC for UPC (≥0.4) (0.7940 vs 0.7241; P = 0.00680). The AUC for the UPC (≥0.4) was not significantly different from that of the AUC for the urine dipstick (≥trace) (0.7241 vs 0.7481; P = 0.4838).

Figure 1.

Receiver operator curve (ROC) for urine protein-to-creatinine ratio (UPC) ≥0.2, UPC ≥0.4, sulfosalicylic acid (SSA) test (≥5 mg/dl) and urine dipstick (≥trace) for the detection of albuminuria as identified by the semiquantitative feline-specific immunoassay (ERD) (≥trace) in the same urine samples

Table 5.

Area under the curve (AUC) for receiver operator curves (ROCs) for the detection of albuminuria as identified by the semiquantitative feline-specific immunoassay (ERD) (≥trace) in the urine samples.

Test	AUC for ROC
Urine dipstick *	0.7481 ‡§
SSA test †	0.7940 *
UPC ≥0.2	0.8315 ‡
UPC ≥0.4	0.7241 §

^†SSA test considered positive when ≥5 mg/dl

^*Urine dipstick was considered positive when ≥trace

^‡§Within column, values with different superscript symbols differ significantly (P <0.0083)

SSA = sulfosalicyclic acid; UPC = urine protein-to-creatinine ratio

Discussion

Traditionally, the urine dipstick and SSA have been used as screening tests and the UPC or, less commonly, the urine albumin to creatinine ratio (UAC) have been used to quantify urine protein concentrations. ⁷ Recently, when compared with a feline-specific quantitative urine albumin immunoassay, the urine dipstick and SSA showed poor specificity (11.0% and 25.4%, respectively) for the detection of albumin in urine of healthy cats. ¹⁰ If the definition of a positive test for the urine dipstick was increased from >trace to ≥2+ the specificity for the dipstick increased from 49.7% to 80.0%, while the NPV decreased slightly (55.0% to 47.3%). ¹⁰ Even when interpreted in series, the urine dipstick and SSA had a low specificity and PPV (26.8% and 47.4%) in healthy cats. ¹⁰ In addition, in healthy cats the UPC was relatively specific, but had a low sensitivity when a cutoff of ≥0.2 or ≥0.4 was used to define a positive test (32.7% and 2.04%, respectively). ¹⁰ Mardell and Sparkes found a significant correlation between the ERD and UPC, although 10/84 (mostly sick) cats that were ERD negative had a UPC of 0.5–1.0. ¹⁵

In the current study, the urine dipstick, SSA and UPC performed better in cats with CKD, as defined by ERD test results than reported previously for healthy cats. ¹⁰ The urine dipstick (≥trace), when interpreted in series with the SSA (≥5 mg/dl), had a high specificity (95.9%) and high PPV (93.2%). A LR+ of 14.5 indicates that if the urine dipstick is ≥trace and the SSA is positive (≥5 mg/dl) then there is very likely albuminuria. In fact, regardless of the urine dipstick, if the SSA is positive then albuminuria is likely (LR+, 15.4). In either case, urine protein quantification via the UPC or a feline-specific quantitative urine albumin immunoassay is warranted if proteinuria/albuminuria is persistent (Figure 2). When the urine dipstick is used alone and ≥trace is considered a positive result, the specificity and PPV are relatively low (68.0% and 70.9%, respectively). When the definition of a positive urine dipstick test is increased to ≥2+ and ≥3+ the specificity is improved (97.5% and 100%, respectively) at the expense of sensitivity (42.7% and 12.0% respectively). A LR+ of 17.4 indicates that a urine dipstick of ≥2+ greatly increases the probability of albuminuria. If the urine dipstick is ≥2+ the SSA adds little to the diagnostic value of a positive urine dipstick result and if the urine dipstick ≥3+ the SSA, as a confirmatory test, is unnecessary. This information indicates collectively that a positive urine dipstick of trace or 1+ should be confirmed with the SSA and that a urine dipstick ≥2+ or a positive SSA (regardless of urine dipstick) are indicative of albuminuria and warrant urine protein quantification if proteinuria/albuminuria is persistent (Figure 2).

Figure 2.

Algorithm for the diagnosis of albuminuria in cats with chronic kidney disease. SSA = sulfosalicylic acid; UPC = urine protein-to-creatinine ratio

The NPV and LR− for the urine dipstick (≥trace) was similar to the NPV and LR− for the urine dipstick (≥trace) interpreted in series with the SSA (≥5 mg/dl) (Table 2). This information would indicate that if the urine dipstick is negative then the SSA, as a confirmatory test, adds little diagnostic information (Figure 2).

In cats with CKD reported here, the UPC performed well when compared with the ERD. A UPC ≥0.4 was always indicative of increased albuminuria. This is in contrast to findings of a previous study which found that 10/84 cats with a negative ERD actually had a UPC of 0.5–1.0. ¹⁵ A UPC ≥0.2 had a high NPV and small LR− (84.6% and 0.19, respectively). This indicates that if the UPC is <0.2, albuminuria is unlikely, and if the UPC is ≥0.4 albuminuria is very likely. There is evidence that a UPC ≥0.2 may be clinically important in cats with CKD.^1,2 The data reported here would support a UPC ≥0.2 as a logical cut-off for a positive test. The combination of the LR+ (4.65) and LR− (0.19) indicate that, with UPC ≥0.2, both a negative and positive test result provide useful information. Using a UPC cutoff of ≥0.2 was also supported by the AUC for the ROC for this test, which was significantly higher than the AUC of UPC (≥0.4). Although the AUC for the UPC (≥0.2) was not significantly higher than the AUC for the SSA or the urine dipstick (≥trace), the higher sensitivity and lower LR− of the UPC (≥0.2) makes this the single best test, especially when minimizing the chance of missing proteinuria is desired.

The reason that the urine dipstick, SSA and UPC performed better in this study compared with previous studies is likely multifactorial. In the two aforementioned studies, cats with CKD were not included in one study ¹⁰ and uncommonly included in the other. ¹⁵ The USG of the urine samples analyzed were not reported in either study, although they would be expected, on average, to be more concentrated than the cats with CKD reported here. Concentrated urine may lead to a false-positive result on the urine dipstick and thus likely led to the lower specificity found for the urine dipstick in healthy cats. ¹⁰ In healthy cats reported previously ¹⁰ the mean (±SD) UPC was 0.14 (±0.017) (unpublished data) and in the cats with CKD reported here the mean UPC was 0.37 (±0.738). It is also possible that a larger fraction of the excreted protein is albumin in cats with CKD. Higher concentrations of albumin or a larger fraction of protein being albumin may have contributed to all tests being more accurate in cats with CKD compared with healthy cats.

This study has several limitations. A semiquantitative feline-specific urine albumin immunoassay (ERD) was used as the ‘gold standard’. This test performed similarly to a quantitative feline-specific immunoassay in a study investigating the ability of both tests to detect systemic disease and both performed better than the urine dipstick and UPC. ¹³ There is no published data, however, comparing the semiquantitative and the quantitative urine albumin immunoassay. Because the ERD requires interpretation of band intensity/color, some subjectivity is involved. Mardell and Sparkes found that when two separate readers examined 26 urine samples on five separate occasions the results were identical 73% and 77% of the time, respectively. ¹⁵ On only two (of 26) occasions was there a discrepancy between a positive and negative test result, as defined in the present study. In addition, in the present study the majority of results were read by one of two people (GFG or SS). On rare occasions, when a test result was difficult to determine, a second opinion was sought. This practice of dichotomizing ERD results as simply positive or negative, having the results interpreted primarily by one of two people and seeking a second opinion if needed, decreased test interpretation variability.

It should be noted that although the ERD is highly sensitive and specific for the detection of albuminuria, the clinical importance of this is still unknown. To date, most studies relating urine protein concentrations to clinical outcome variables in cats with CKD have used the UPC or UAC.^1–5 As stated previously, the ERD performed better when compared with the UPC and urine dipstick for the diagnosis of systemic disease. ¹³

In conclusion, the commonly used screening and confirmatory tests for the detection of albuminuria performed better in cats with CKD compared with that reported previously for healthy cats. ¹⁰ In cats with CKD, a trace or 1+ dipstick and positive SSA, positive SSA alone (regardless of urine dipstick) or a positive urine dipstick (≥2+) alone are indicative of albuminuria and warrant urine protein quantification if proteinuria is persistent. In the case of a negative urine dipstick, albuminuria is unlikely and, unless otherwise warranted, further investigation is likely unnecessary. Of the tests investigated, a UPC ≥0.2 was found to be the single best test for the detection of albuminuria in cats with CKD.