Leukocyte Esterase and Nitrite Urine Reagent Strip Utility Under Altered Assay Conditions in Dogs
ABSTRACT
The study evaluated leukocyte esterase (LE) and nitrite reagent pad usefulness in canine urine using multiple time endpoints and decision thresholds. Whole and sedimented urine samples from 116 client-owned dogs were analyzed with reagent strips from four manufacturers. Results for LE and nitrite were recorded every 30 s up to 150 s using “trace” and “+” positive thresholds and compared to microscopic urine sediment analysis and culture results. Sedimented urine assessed at 150 s with a “trace” positive threshold for LE had significantly or trending higher sensitivity (brand dependent) for pyuria detection compared with manufacturer recommendations. Using these parameters, LE sensitivity and negative likelihood ratio were 66.67–89.47%, and 0.13–0.37, respectively. Following manufacturer recommendations, LE specificity and positive likelihood ratio were 96.91–100% and 17.02–30.63, respectively. Nitrite performance for pyuria and bacteriuria detection was poor for all conditions. LE reagent pads are a useful screening test for pyuria detection using sedimented urine and a “trace” positive threshold at 150 s, whereas LE use with traditional manufacturer recommendations is a good confirmatory test for pyuria detection. Nitrite reagent pads are not useful for pyuria or bacteriuria detection.
Introduction
Urinalysis is a fundamental component of a minimum database in small animal medicine. A standard, complete urinalysis is comprised of evaluations of chemical and physical properties, solute concentrations, and a microscopic evaluation of the urine sediment.1 A common urinalysis indication is for the detection of pyuria in an animal exhibiting clinical signs concerning for a urinary tract infection (UTI). Pyuria is defined as the presence of leukocytes in urine. However, reference intervals vary by laboratory. Other factors affect the interpretation of urine leukocyte concentration, such as volume or rate of urine produced, urine specific gravity, time of collection, sediment preparation, medications, and health conditions, which may influence the inflammatory response.1
Many commercially available urine reagent strips, which are manufactured for use with human samples but widely used in veterinary medicine, contain leukocyte esterase (LE) and nitrite reagent pads. The LE reagent pad contains an ester substrate, which varies by manufacturer. Human neutrophil primary granules contain up to 10 enzymes with ester hydrolyzing properties.2 According to the manufacturersa,b,c,d, ester hydrolysis results in the production of an intermediate compound (also varies by manufacturer), which reacts with a diazonium salt, resulting in a purple color change, indicative of the presence of pyuria. The nitrite reagent pad is used for the detection of bacteria that can reduce urine nitrate into nitrite. The resulting diazonium compound from a reaction with nitrite and the reagent pad aromatic amine (often para-arsanilic acid) then reacts with another compound (varies by manufacturer) to yield a pink color changea,b,c,d. In human medicine, urine reagent strips are used routinely as a diagnostic tool when assessing patients with signs potentially attributable to UTIs. LE results, in combination with those from the nitrite reagent pad, provide a sensitive, accurate assessment for the presence of a UTI.3 Meta-analyses have suggested that using these reagent strip results may be appropriate for ruling out UTIs, with a need to consider clinical information that may affect pretest probability.4,5
Results from previous studies have suggested that LE reagent pads have little to no clinical utility in dogs and cats. An early study showed that detection of pyuria in dogs by the LE pads had a sensitivity and specificity of 46% and 93.2%, respectively.6 Another study investigating the detection of pyuria in dogs showed poor correlation between LE reagent pad results compared to urine sediment microscopic evaluation, with a sensitivity of 9.6% and a specificity of 100%.7 Postulations for the cause of low sensitivity in canines have included the presence of urinary enzyme and/or esterase inhibitors and differences in enzyme quantity or activity compared to humans.6 Unpublished studies from the University of Minnesota found the sensitivity of the LE reagent pad to be 75 leukocytes per microliter; however, the specifics of the study are not available.1 Another unpublished canine study from the University of Minnesota and the University of Guelph found that the LE reagent pads are highly specific (100%) but insensitive (62%) for detecting bacteriuria, based on urine culture results.1 Conversely, the detection of pyuria by LE pads in cats had a specificity of 34% and a sensitivity of 77%.8 The authors’ considerations for the high rate of false positives in cats included neutrophil lysis, increased quantity and/or activity of feline esterases (compared to humans) and possible cross-reaction of nonleukocyte esterases or oxidizing agents. The protein cauxin, a carboxylesterase, is known to cause false positives with reagent strip analysis for protein and has been separately proposed as a reason for false positive results with reagent strip results for LE.9 Although little published research on nitrite reagent pads in veterinary literature exists, it is widely acknowledged that these reagent pads are unsuitable for canine and feline urine analysis.1,10–12 The inhibitory effects of ascorbic acid on the assay in canine and feline urine, the types of bacterial species that commonly affect domestic animals, and insufficient time for bacterial reduction of nitrate to nitrite have been proposed as possible causes for high numbers of false negatives, contributing to the unsuitability of this assay in canines.
Although the routine use of many reagent pads on commercially available urine reagent strips is widespread in veterinary medicine, their use in animal samples is off-label. Prior study designs aimed at investigating the utility of LE and nitrite reagent pads on urine reagent strips have followed the manufacturers’ instructions, which were created and optimized for human samples. Previous studies with canine and feline samples have demonstrated that there are significant species differences that preclude the use of the LE and nitrite reagent pads with canine and feline urine samples according to manufacturers’ recommendations.6–8,10 The aim of the present study is to investigate the analytical conditions under which these pads may have clinical utility. Our objective was to evaluate the diagnostic capability of LE and nitrite reagent pads from multiple manufacturers with urine sediment samples and whole-urine samples measured at different time points and assessed with different semiquantitative decision thresholds. To the authors’ knowledge, veterinary studies have not investigated altered diagnostic parameters to optimize these reagent pad assays for use in veterinary species.
Materials and Methods
Obtaining Urine Specimens and Inclusion Criteria
This prospective study was performed between January and November 2020. Convenience sampling of all canine urine specimens of at least 3 mL that were submitted to the Veterinary Teaching Hospital at the University of Georgia were included in the study. All methods of urine collection were accepted. Urine samples that were moderately to markedly discolored, precluding assessment by reagent strip pads, were excluded.
Urinalysis and Sediment Preparation
Routine urinalyses (reagent strip evaluation, urine specific gravity via refractometry, and sediment wet mount microscopy) were performed as soon after urine collection as possible, within 2 hr. Following gentle inversion or swirling, a urine reagent stripe was immersed in whole urine and removed immediately. Results from the pH, protein, blood, ketones, bilirubin, and glucose reagent pads were recorded according to manufacturer instructions. Urine specific gravity was measured via refractometryf. The refractometer was calibrated daily with distilled water. For urine sediment sample preparation, 2.5 mL of whole urine was centrifugedg in a 12 × 75 mm tube at 2400 rpm for 5 min. After sedimentation, the pellet was resuspended in 250 μL of reserved supernatant to create the urine sediment, and excess supernatant was discarded. One drop (approximately 30 μL) of urine sediment was placed on a microscope slide using a transfer pipette. A 22 × 22 mm glass coverslip was placed on top of urine sediment drop for microscopic evaluation. White blood cells were enumerated over ten FN22/400X microscopic fields and reported as an average number per 0.237 mm2 (equivalent to one FN22/400X microscopic field).
LE and Nitrite Reagent Pads
For each urine sample, approximately 30 μL drop of whole urine was placed on LE and nitrite reagent pads using a transfer pipette and results were recorded at 30, 60, 90, 120, and 150 s. Separately, this process was repeated with urine sediment for each urine sample at each time point. For LE and nitrite measurements, reagent strips from four different manufacturersa,b,c,d (products will be subsequently designated brand Ac, brand Ba, brand Cd and brand Db) were used for each urine sample. Results were recorded as negative, trace, +, ++, or +++. Manufacturer recommendations for the use of LE reagent pads include the use of whole urine, and brands B and C provide a range of reaction times, but the 120 s time point is common among all brands. Positive thresholds are variably addressed by urine reagent strip package inserts. Brand A refers to a positive reaction as “+” or greater, and brands A, B, and D suggest that a “trace” result is a borderline situation or of questionable clinical significance. As such, for LE reagent pads, whole-urine samples assessed at 120 s with a “+” threshold for a positive result will subsequently be referred to as manufacturer recommendations.
Ancillary Testing
If a concurrent urine culture was submitted, the results were recorded. As available, other urine diagnostic results such as urine protein to creatinine ratio were also recorded.
Statistical Analysis
LE and nitrite reagent pad results were independently compared to the presence or absence of pyuria based on urine sediment microscopic examination as well urine culture results. Pyuria was diagnosed when urine sediment samples contained an average of ≥5 leukocytes per 0.237 mm2 (equivalent to one FN22/400X microscopic field). Sensitivity, specificity, positive likelihood ratio (LR+), and negative likelihood ratio (LR−) were used to assess diagnostic performance of the LE and nitrite reagent pads compared to microscopic examination (presence of pyuria) and urine culture (presence of bacteriuria). These parameters were calculated for each set of unique conditions with combinations of four independent variables: reagent strip brand, time (30, 60, 90, 120, or 150 s), sample type (whole urine or sediment), and threshold interpretation of a positive result (“+” or “trace”). For each diagnostic parameter, 95% confidence intervals (CIs) were calculated, and significant differences between conditions or reagent strip brands were designated by the presence of nonoverlapping CIs.
Results
Samples
Urine samples from 116 dogs met the inclusion criteria during the study period and were included for analysis. Because of the unavailability of brand D reagent strips during part of the sample collection period, 102 urine samples were analyzed for this brand whereas the brands A–C included all 116 urine samples. The mean age of included dogs was 9.1 yr with a range from 4 mo to 18 yr. Fifty-six dogs were spayed females, 47 were castrated males, 8 were intact males, and 5 were intact females. Fifty-four breeds were included with 23 (19.8%) mixed-breed dogs, 8 (6.9%) German shepherd dogs, and 6 (5.2%) shih tzus representing the most common breeds. Pyuria was identified in 19 samples, yielding an overall 16.38% prevalence rate. Urine culture was performed on 29 samples, 15 (52%) of which were positive. Twelve of the 19 samples that had detectable pyuria based on microscopic examination were cultured, and of those 12 samples, 10 exhibited bacterial growth under culture conditions. Of the 15 cases that were cultured positive, 10 had microscopic pyuria and 5 did not, whereas of the 14 cases that were culture negative, 2 had pyuria and 12 did not. Cultured organisms included Escherichia coli (7), Proteus mirabilis (2), Streptococcus infantarius, Citrobacter braakii, Enterobacter hormaechei, Bacillus pumilus/altitudinis, Klebsiella pneumoniae, and another Bacillus spp.
Comparison of LE Reagent Pad Results with Urine Sediment Microscopic Examination
Urine sediment samples assessed at 150 s with a “trace” positive threshold provided the highest sensitivity for all brands, ranging from 66.67% (95% CI 43.75–83.72) (brand D) to 89.47% (95% CI 68.61–97.06) (brand B) (Figures 1 and 2, Table 1). Sensitivity for urine sediment samples assessed at 150 s with a “trace” positive threshold was significantly increased compared with that of whole-urine samples assessed at 120 s with a “+” positive threshold (manufacturer recommendation) for brand A reagent strips. Although not statistically significant, sensitivity increases at 150 s relative to 120 s and, when using a “trace” rather than “+” positive threshold, were strong trends among the three other brands.
Citation: Journal of the American Animal Hospital Association 58, 5; 10.5326/JAAHA-MS-7233
Citation: Journal of the American Animal Hospital Association 58, 5; 10.5326/JAAHA-MS-7233
Urine sediment samples assessed at 150 s with a “trace” positive threshold also provided the lowest specificity for all brands, ranging from 80.41% (95% CI 71.42–87.09) (brand B) to 89.29% (95% CI 80.88–94.26) (brand D) (Figures 1 and 2, Table 2). Specificity for sedimented urine was significantly decreased compared with that of whole-urine samples assessed at 120 and 150 s with a “+” positive threshold across all brands.
High LR+ in whole-urine samples assessed at 120 or 150 s with a “+” positive threshold ranged from 10.21 (95% CI 3.93–26.51) (brand B) to 35.74 (95% CI 4.66–273.95) (brand A) (Figures 1 and 2). LR+ were not available for brand D with these assay conditions because of zero false positives. In contrast, sediment samples assessed at 150 s with a “trace” positive threshold provided the lowest LR+, ranging from 4.57 (95% CI 2.97–7.03) (brand B) to 6.38 (95% CI 3.58–11.38) (brand A).
The lowest LR- were found in urine sediment samples assessed at 120 or 150 s with a “+” or “trace” positive threshold, depending on the brand of urine reagent strip (range from 0.12 [95% CI 0.03–0.46] [brand B] to 0.37 [95% CI 0.19–0.72] [brand D]) (Figures 1 and 2).
Statistically significant differences for sensitivity, specificity, LR+, or LR− between brands were not found. However, several trends were observed, including higher sensitivity with brand B with all conditions, lower sensitivity with brand D with most conditions, and lower specificity with brand B with most conditions.
Comparison of LE Reagent Pad Results with Urine Culture
Similar to the LE comparison with urine sediment microscopic examination, urine sediment samples assessed at 150 s with a “trace” positive threshold provided the highest sensitivity for all brands, ranging from 71.43% (95% CI 45.34–88.28) (brand D) to 73.33% (95% CI 48.05–89.10) (brands A, B, and C). Sensitivity using these conditions was significantly increased compared with that of manufacturer recommended conditions for brand A. The sensitivity increase between these assay conditions was a strong trend, rather than statistically significant, among other brands.
Specificity was slightly to mildly decreased for urine sediments when assessed at 150 s with a “trace” positive threshold compared with manufacturer recommended conditions for all brands; however, these were not statistically significant.
Comparison of Nitrite Reagent Pad Results with Urine Sediment Microscopic Examination
All assay conditions for all brands exhibited low sensitivity for the detection of pyuria. Although high specificities are reported, the number of overall positive results (false positives plus true positives) ranged from 2 to 8. No significant differences in sensitivity or specificity were observed between brands.
Comparison of Nitrite Reagent Pad Results with Urine Culture
Similar to the comparison of nitrite reagent pads to urine sediment microscopic examination, sensitivity for predicting a positive culture result was low for all brands at all conditions. High specificities are reported, but the number of overall positive results (false positives plus true positives) ranged from 0 to 3. No significant differences in sensitivity or specificity were observed between brands.
Discussion
A primary aim of the study was to investigate whether alterations of assay conditions may increase the clinical utility of LE and nitrite reagent pads on commercially available reagent strips for canine urine, specifically longer exposure and sedimented versus whole urine. LE reagent pads have been shown to be specific but poorly sensitive for detecting pyuria in canine urine samples using standard human manufacturer recommendations.6,7
The results from the current study also showed poor sensitivity and high specificity for the detection of pyuria by LE reagent pads using manufacturer recommended assay conditions (Figures 1 and 2, Tables 1 and 2). Because poor sensitivity characterized the poor performance and accuracy of these reagent pads in prior studies, this study focused on aspects that would increase sensitivity for pyuria detection. Alterations of independent variables deviating from manufacturer recommendations for human samples (increasing the time between application of sample on the reagent pad to assessment of result from 120 to 150 s, using urine sediment rather than whole urine, and using a threshold result of “trace” rather than “+”) each increased this sensitivity, although not significantly, when assessed individually. When the combination of these three alterations were applied to the assay conditions, there was either a significant increase in sensitivity or a strong trend that approached a significant increase depending on the brand of urine reagent strip (Figures 1 and 2, Table 1). This increase in sensitivity accompanied an expected decrease in specificity; however, the LE reagent pads from all brands still had specificities above 80% (Figures 1 and 2, Table 2).
Despite high specificities reported in prior studies, the use of the LE reagent pad is not reported in referral clinical pathology laboratories, and its clinical use is not generally adopted. Results from prior studies as well as the current study support that the use of the LE reagent pad according to manufacturer recommendations has a high specificity for the detection of pyuria. As such, although a negative result using these assay conditions may be clinically unhelpful, a positive result may be highly valuable as a predictor for the presence of pyuria. The LR+ for LE reagent pads using manufacturer recommendations for pyuria detection confirms the utility of this assay.
Likelihood ratios provide the clinician with a single value that represents the probability of an affected patient (a sample with pyuria, in this instance) receiving a particular test result (positive or negative) divided by the probability that an unaffected patient (a sample without pyuria) receives the same result.13 LR+ for all brands well exceed 10, which is a generally accepted benchmark for a useful test when ruling in a condition.13 As an example to illustrate the clinical utility of this ratio, the highest calculated LR+ in this study using LE reagent pads at manufacturer recommendation to detect pyuria (brand A) indicates that urine samples with pyuria are approximately 30 times more likely to test positive than samples without pyuria. Of note, the LR+ calculation for brand D was not possible because of the lack of false positives, resulting in an LR+ equation with a denominator of zero. The lack of a valid LR+ in this case suggests a very high LR+. The performance of the LE assay, following manufacturer recommendations, fits the profile of a good confirmatory test but a poor screening test for pyuria. It can be performed as part of the routine reagent strip urine analysis, adding no additional cost, negligible extra sample volume, and minimal extra assay time.
By changing assay conditions using urine sediment samples assessed at 150 s and a “trace” threshold for a positive result on the LE pad, it appears to be a useful screening tool for pyuria. This may be especially true for brand B, which showed the highest sensitivity at 89.47% (95% CI 68.61–97.06) for detecting pyuria using these conditions (Figures 1 and 2, Table 1). A positive result while using this altered condition LE assay may provide good clinical justification for the recommendation for follow-up testing, such as a send-out urinalysis and/or a urine culture. This is further supported by the increased sensitivity of LE reagent pads compared with urine culture results when used under these conditions.
With relatively few false negatives contributing to a low LR−, a negative result using these conditions can provide quick, inexpensive support for the absence of pyuria (Figure 1). The lowest LR− was 0.13 using the altered conditions with brand B, reflecting that a sample tested without pyuria is 7.7 times more likely to yield a negative result than a sample with pyuria. Ideal negative predictive values are generally 0.1 or less for ruling out disease; however, 0.13 approaches this mark and, as such, may have clinical utility, especially when interpreted concordantly other clinical data.13
Using the LE reagent pad according to manufacturer recommendations as a confirmatory test in combination with the altered protocol as a screening test may be particularly useful in situations when limited financial resources may dictate clinical decision making. The use of this altered protocol may be particularly suitable when canine urine is being analyzed only within the clinic without a microscopic examination from dogs with a lack of urinary clinical signs, especially if there are predisposing conditions (i.e., diabetes mellitus).
The nitrite reagent pad is widely accepted to be unsuitable for use in dogs and cats.1,10–12 The results from the current study agree with these assessments with very low numbers of total positive test results, particularly true positive test results, and high numbers of false negatives. These both contribute to very low sensitivities when comparing nitrite reagent strip results with urine sediment microscopic examination and with urine culture results. Although high specificities are reported, the very low number of positive assay results across all brands precludes adequate assessment of this assay’s ability to distinguish false positives from true negatives. Many more samples would need to be included for a more accurate evaluation of specificity.
Consideration of methodology is important when evaluating prior studies, as well as when processing and interpreting clinical samples. Urine sediment methodology, in particular, has been variable across prior studies and remains inconsistent among clinics, hospitals, and laboratories. The preparation of urine sediments in the current study was based on the recommendation set forth in a widely used veterinary reference for urinalysis in veterinary patients with mild modification.1 This reference recommends centrifugation of 5 mL whole urine and that 0.5 mL of the urine sample should remain to create the sediment sample, following centrifugation and supernatant removal. To maintain the ratio between starting whole-urine volume to final sediment volume while allowing for greater numbers of cases to be included, 2.5 mL whole urine was used as a starting volume and 0.25 mL final volume for urine sediment.
A feline study showed that the LE reagent pads yield moderately sensitive results (77%) but have poor specificity (34%) when compared with microscopic sediment analysis.8 Preliminary data collected during the current study in 12 cats (data not shown) revealed a nearly uniform, immediate, strong positive reaction of the LE reagent pads upon application of either whole urine or urine sediment. Only 1 cat had evidence of pyuria based on microscopic examination, and only 1 of the 12 cats had a culture submitted. Despite the small sample size, the largely indiscriminately positive preliminary results support that LE reagent pads have no clinical utility in cats despite assay conditions.
Although statistically significant differences were not found between brands, there were some appreciable trends. Variation in detection limit and analytical sensitivity as well as varying reactions to each ester substrate may be considered as possibilities for these differences.
Limitations to consider with the current study include relatively wide CIs and a lack of interference consideration. Some significant changes and some strong trends were observed between different conditions in this study; however, a study with higher numbers of samples could narrow the associated CIs and may improve clinical recommendations. Information regarding presenting complaints, physical examination findings, confirmed or suspected diagnoses, medications, and other possible interferences can affect pretest probabilities for the presence of pyuria and bacteriuria. As such, future studies incorporating these clinical features may help to further characterize the utility of these assays. Presented statistical results were limited to sensitivity, specificity, and likelihood ratios, which are focused on the performance inherent to the assay.
Conclusion
Although urine sediment microscopic analysis remains the gold standard for the determination of the presence of pyuria, inconsistent personnel training across clinical institutions, variable and/or lack of standardized sediment preparation protocols, and variability of access to and type of microscope equipment may all contribute to the appeal of a less variable, more accessible, and concurrently inexpensive assay. The present study supports that LE reagent pad has potential utility as a confirmatory test for pyuria when manufacturer recommendations are applied, as well as a screening test for pyuria when urine sediment is analyzed at 150 s with a “trace” positive threshold. However, nitrite reagent pads are not useful despite variously altered assay conditions with canine urine samples.
Comparison of the binary results (positive or negative) from four leukocyte esterase reagent strip brands, using urine sediment microscopic examination as the gold standard. The descriptors shown on the left designate the assay conditions and threshold at which the reagent strip results are considered positive for pyuria. Either whole urine or sedimented urine was applied to the reagent pad for each assay. Results were assessed at either 120 s (120s) or 150 s (150s) following urine application to the reagent pad. Reagent pad semiquantitative results equal to or greater than either “+” or “trace” were considered positive. For Brands A, B, and C, 116 canine urine samples were used; 102 samples were used for Brand D.
Comparison of the binary results (positive or negative) for sensitivity and specificity of four leukocyte esterase reagent strip brands, using urine sediment microscopic examination as the gold standard. The descriptors found in the legend within the figure designate the assay conditions and threshold at which the reagent strip result is considered positive for pyuria. Either whole urine or sedimented urine was applied to the reagent pad for each assay. Results were assessed at either 120 s (120s) or 150 s (150s) following urine application to the reagent pad. Reagent pad semiquantitative results equal to or greater than either “+” or “trace” were considered positive. For Brands A, B, and C, 116 canine urine samples were used; 102 samples were used for Brand D. The vertical bars reflect the 95% confidence interval for the sensitivity or specificity for the respective assay conditions. Horizontal black lines within these bars represent the respective sensitivity or specificity.
Contributor Notes
