Materials and Methods

Selection of cats—Six hundred eleven cats that were evaluated serially at the Veterinary Medical Center at Colorado State University from March 3, 2003, to May 6, 2005, and for which a urine sample was available were initially evaluated. After elimination of repeated submissions from the same cat and cats without a complete medical record, data on 441 cats remained. The medical record for each cat was reviewed by 1 of 2 authors (JCW or ZM) who were blinded with regard to urinalysis results at the time of data collection. Rectal temperature, systolic blood pressure, diagnostic tests performed, current medications, and clinical diagnoses entered within 3 months of the time of urine collection were recorded. Clinical diagnoses included healthy; and neoplastic; infectious, inflammatory, or immune-mediated; urinary and renal; endocrine; and other diseases.

Urine assays—Urine samples collected by free catch, catheterization, or cystocentesis were included. Urinalyses were performed by technicians in the clinical pathology laboratory at the veterinary teaching hospital, and samples were frozen at −20°C until the other assays were performed. Technicians blinded to the urinalysis results performed the MALB_E; samples were refrozen for transport and submitted to an outside laboratory^f where the MALB_Q, urine total protein assay, and urine creatinine assay were performed by technicians blinded to the urinalysis results. The UPC and UAC ratios were calculated by dividing the total protein or quantitative microalbuminuria concentration by the creatinine concentration in each sample prior to standardization for specific gravity. All assays were performed according to methods that have been described.⁵

Statistical analysis—Urine protein dipstick results of trace or greater were considered positive. Microalbuminuria test results (MALB_E or MALB_Q) ≥ 1 mg/dL were considered positive. Urine albumin-to-creatinine ratio cutoff values of 100 (UAC₁₀₀) and 200 (UAC₂₀₀) were evaluated; these values were chosen on the basis of data from humans and cats.¹ Urine protein-to-creatinine ratio cutoff values ≥ 0.1 (UPC_0.1) and ≥ 0.4 (UPC_0.4) were evaluated. Diagnostic sensitivity and specificity of the MALB_Q and MALB_E, UPC_0.1 and UPC_0.4 ratios, UAC₁₀₀ and UAC₂₀₀ ratios, and dipstick protein results were calculated by use of χ² analysis, with the presence or absence of systemic disease designated as the independent variable. Stepwise backward-selection logistic regression was performed to evaluate the association between disease status, sex, age, BUN, serum creatinine concentration, results of urine bacterial culture, systolic blood pressure, rectal temperature, pyuria, hematuria, or bacteriuria and results of the MALB_Q, MALB_E, or dipstick protein tests. Logistic regression was performed on subpopulations of various disease categories. Results of subpopulation evaluations where model-fitting procedures converged were summarized. Values of P < 0.05 were considered significant. Odds ratios and confidence intervals were determined where possible. All statistical analyses were performed according to previously described methods by use of commercially available software.^g

View Full Size

Distribution of MALB_Q values by diagnosis category in 441 cats with or without disease. Values of MALB > 50 mg/dL (n=3cats)havebeen removed for clarity (2 cats with urinary tract disease and 1 cat with multiple diseases). Healthy = Cats that had no disease conditions. Disease = All cats not categorized as healthy. Endocrine = Cats with endocrinologic disease. Inf = Cats with infectious, immune-mediated, or inflammatory disease. Mult = Cats with > 1 diagnosis. Neop = Cats with neoplastic disease. Other = Cats with 1 disease that did not fall into one of the other primary categories. Urinary = Cats with urinary tract disease. The dotted line represents the cutoff value of 1 mg/dL. The line in each box is the median MALB_Q value for that diagnosis category; boxes encompass the 25th to 75th percentiles, whiskers are the 10th and 90th percentiles, and individual dots represent outliers.

Citation: Journal of the American Veterinary Medical Association 230, 8; 10.2460/javma.230.8.1165

• Download Figure
• Download figure as PowerPoint slide

Results

Of 441 cats for which medical records were available and included in analyses, > 1 disease process was identified in 117. Of the remaining 324 cats, 40 (12%) were classified as healthy; 53 (16%) had neoplasia; 52 (16%) had infectious, inflammatory, or immune-mediated diseases; 61 (19%) had urinary tract disease; 52 (16%) had endocrine disease; and 66 (20%) had other diseases (Table 1). The distribution of microalbuminuria values among the diagnosis categories was not the same (Figure 1). Four urine samples were obtained from indwelling urinary catheters, and the other samples were obtained by either free catch or cystocentesis. Distribution of cats with positive results via MALB_Q and MALB_E, UPC_0.1 and UPC_0.4, UAC₁₀₀ and UAC₂₀₀, and dipstick protein by diagnostic code was summarized. The sensitivity and specificity of the MALB_Q and MALB_E, UPC_0.1 and UPC_0.4, UAC₁₀₀ and UAC₂₀₀, and dipstick protein for distinguishing healthy from nonhealthy cats were determined (Table 2). Results calculated with and without cats with multiple diagnoses were not significantly different.

All cats had positive results when the 0.1 cutoff value for UPC ratio was used, so the UPC_0.1 test was not evaluated further. The small number of cats with positive results via UPC_0.4 (n = 16), UAC₁₀₀ (16), and UAC₂₀₀ (7) precluded statistical evaluation of those tests. The number of cats with available blood pressure or rectal temperature measurements was insufficient to assess for association with urine test results. Factors significantly associated with positive MALB_Q results were health status, presence of urinary disease, azotemia (BUN or serum creatinine concentration, depending on subgroup), pyuria, and hematuria. Factors significantly associated with positive MALB_E results were health status, presence of urinary disease, age, azotemia (BUN or serum creatinine concentration, depending on subgroup), pyuria, and hematuria. The only factor significantly associated with dipstick protein status was hematuria. Neoplasia and urinary tract disease were the only disease subcategories for which there were enough cats to perform subgroup regression analysis. Significant associations were summarized (Table 3).

Table 3—

Results of logistic regression analysis for associations between positive microalbuminuria assay or dipstick protein results and disease status in cats.

View Table

Discussion

Proteinuria may be classified as prerenal, functional renal, pathologic tubular, or pathologic glomerular in etiology. Overt proteinuria (UPC ratio > 1.0) is commonly accepted as abnormal and generally indicates a need for further diagnostic testing. Data from studies in dogs,^5,h cats,^1,a,i and humans^6,7 suggest that subtle proteinuria, manifested as microalbuminuria, may be an indication for further evaluation because it may be associated with systemic disease, morbidity, and higher all-cause mortality rates.

Given the low specificity and sensitivity, respectively, of UPC_0.1 and UPC_0.4, these indices appear to have limited value for screening cats without overt proteinuria for systemic disease. The low sensitivity of the UPC ratio with a cutoff of 0.4 was consistent with results of recent studies^1,a and suggests that an even higher cutoff for a positive result of 1.0, as is currently used in practice, may be too high for use in cats. The low prevalence of proteinuria, determined on the basis of a UPC ratio > 0.4, in this and the previously cited studies,^1,a contrasts markedly with the prevalences of proteinuria (defined by a UPC ratio > 0.5) of 34% in diseased cats and 5% in healthy cats reported in another recent study.⁸ Differences in study population, the prevalences of various disease categories, and the methodologies used to measure the UPC ratio may be associated with the differing results.

The absence of association between dipstick protein results and pyuria was unexpected given results of a previous report² on the specificity of the MALB assays versus dipstick testing for diagnosis of urinary tract inflammation. However, to date, most animals with pyuria that were assessed with microalbuminuria assays and dipstick protein analysis were dogs. Although the explanation for this discrepancy between cats and dogs is unknown, our results suggest that in cats, false-positive dipstick protein results associated with pyuria are unlikely to occur.

The poor usefulness of UAC₁₀₀ and UAC₂₀₀ ratios for screening cats for systemic disease was unexpected. Previous work in cats revealed a significant association between the UAC ratio and hypertension and azotemia.¹ The UAC ratio is also a predictor of fatality in apparently healthy cats as well as cats with those conditions.^1,a There are several possible explanations for this apparent conflict in results. It may be that the UAC ratio is not sensitive enough to function as a marker for disease despite the association between increasing values of UAC and disease. In the previously mentioned studies,^1,a an endpoint of death and a repeated sampling technique were used, whereas in the present study, cats were evaluated for only 3 months and single time point sampling was used. It may be that the shorter time frame and single time point sampling used in the present study were not sufficient to evaluate this relationship. Although differences in the study population may have affected the results, we would have anticipated a higher prevalence of cats with positive results on the basis of the UAC ratio in this population because of the small number of healthy cats. Urine handling and storage conditions were not different among the various studies and should not have affected recovery of albumin from urine. Finally, sampling bias cannot be ruled out as a factor. Clinicians may have had more difficulty obtaining urine from cats with more severe illness, resulting in a bias toward sampling more stable sick cats with less severe disease.

The differences in the association between explanatory variables and individual microalbuminuria tests were not unexpected given the potential differences in the performance of these 2 tests. The association between BUN, serum creatinine concentration, and presence of urinary disease and positive microalbuminuria results was not surprising and was consistent with findings from a previous study.¹ The lack of a sufficient number of cats to analyze for an association between microalbuminuria and hypertension was unfortunate because results are conflicting in current scientific literature.^1,8 The absence of an association between microalbuminuria and neoplasia differed from the results we obtained in dogs, where the odds ratios for neoplasia were 2.7 and 8 for positive MALB_Q and MALB_E results, respectively.⁵ A potential explanation is that the neoplasms in cats of the present study were less inflammatory than those affecting dogs. We consider this to be unlikely because a large proportion (17/53 [32%]) of the cats with neoplasia in the present study had vaccine-associated sarcomas, which are considered to be highly inflammatory.⁹ Alternatively, immune complexes and other inflammatory mediators may interact differently in dog and cat glomeruli. Unfortunately, because only those 2 categories had a sufficient number of cats for the statistical model to converge, further conclusions about associations between specific disease categories or types of disease and proteinuria cannot be drawn.

This study had a number of limitations. To prevent potential bias or operator variation, samples were analyzed en masse. This approach helped limit diagnostic workup bias because urine test results were not available to the attending clinicians involved with the cats. The disadvantage of this approach was that clinicians did not have the opportunity to use urine test results in clinical evaluation of cats. Because this study was observational in nature, another important limitation was that cats with ongoing disease were included in study enrollment. Some of these cats were in remission for incurable disease, and some were nearly recovered from previously identified disease. This may limit application of these results to a naïve population. Given the high prevalence of systemic disease in geriatric cats and the continuum of disease in cats when initially evaluated, we feel that this concern has limited validity. Sequential enrollment of cats and the disease characteristics of cats evaluated at a tertiary care facility led to disparity between the number of healthy and unhealthy cats and the relative prevalence of certain conditions. This distribution inequality may have skewed interpretation of the statistical results because the number of positive results was not similar to the number of negative results. Data generated early in the study indicated that degradation of albumin secondary to repeated freeze-thaw cycles affected approximately 10% of urine samples. To avoid introduction of bias, no change was made in the urine-handling protocol. It is possible that degradation of albumin in some samples may have negatively affected sensitivity, specificity, and regression analysis results for the MALB_E, MALB_Q, UAC₁₀₀, and UAC₂₀₀ tests. Finally, inclusion of cats with diseases not known to be associated with proteinuria (eg, dermatologic disease) in the diseased category for logistic regression analysis may have biased the analysis against identifying significant correlations. Such cats were included because other options (censoring these cats or categorizing them as healthy) would have decreased applicability of the study results to general practice populations. In addition, such choices would have required prospective judgment of which diseases should or should not be associated with microalbuminuria without the benefit of objective supporting data.

In the present study, UAC ratios were not useful in identifying underlying disease because of poor sensitivity. In contrast, 1 or both of the MALB assays was associated with the presence of underlying disease, the presence of renal disease, and age. Conflicting results relative to previously reported studies may reflect the complex and multifactorial nature of processes that cause microalbuminuria. In the present study, sensitivity and specificity of the microalbuminuria assays for detection of systemic disease were superior to those of the other tests. It appears that there is benefit for the use of the microalbuminuria assays in conjunction with other screening tests (eg, signalment, history, physical examination, CBC, serum biochemical analysis, urinalysis, and blood pressure) to increase detection of occult disease. Further prospective studies in which the positive and negative predictive values of other screening tests are evaluated, with and without a microalbuminuria assay, are needed to further validate this recommendation.