Basal glucose excretion in dogs: The impact of feeding, obesity, sex, and age

Abstract Background The urine glucose (UG) measurements are an integral part of urinalyses, especially in dogs with polyuria and polydipsia. A positive dipstick result is considered pathologic for disease. This paradigm has been challenged by new ultrasensitive tests, where the manufacturers recommend tolerating slightly positive results. It implies that, as in other species, basal urine glucose losses can exceed the lower limits of detection using ultrasensitive glucose dipsticks in healthy dogs. Objectives We aimed to determine whether glucose is routinely detectable using a sensitive quantitative wet chemistry method in the urine of nondiabetic, nonazotemic dogs, and investigate the impact of food intake, obesity, sex, castration status, and age. Methods Serial UG measurements were performed in healthy clinic‐owned Beagle dogs that were randomly fasted or fed. Glucose was measured in morning urine samples from normal‐weight healthy and obese dogs, and the university's electronic database was searched for quantitative UG measurements (Gluco‐quant Enzyme Kit/Roche Diagnostics). Results Small amounts of glucose were detected in 555 (99.1%) of 560 urine samples analyzed. All urine samples from the clinic‐owned Beagle dogs, as well as from privately owned obese and normal‐weight healthy dogs that tested positive for glucose. The median (range) UG concentration obtained from the university's electronic database was 0.39 (0‐1.55) mmol/L, and 2.2% of the samples tested negative. Feeding, obesity, gender, castration status, and age did not affect UG concentrations. Conclusions Studies, including a larger number of healthy dogs, are warranted to define a cut‐off between physiologic and pathologic glucosuria.

False-positive dry reagent strip results ("pseudoglucosuria") can be caused by antibiotics such as cephalexin, 11 contamination by disinfectants such as hydrogen peroxide, 12 and prolonged exposure of the reagent strips to air. 13 It is noteworthy that experimental blood contamination caused by adding euglycemic blood samples onto strips had no significant effect on UG scores. 14 The use of automated dipstick readers can reduce the error rates associated with dipstick urinalyses. 15 A recent study found discordant results between urine dip and urine drip method, with more trace positive glucose results in the urine of nondiabetic dogs without evidence of tubular disease when using the drip method. 16 The lowest UG concentration estimate using traditional test strips was 2.8 mmol/L (50 mg/dL, eg, Combur 9; Roche Diagnostics) The objectives of this study were to determine whether basal glucose excretion is habitually detectable in the urine of nondiabetic dogs when analyzed by a sensitive automated wet chemistry method. The scientific background was, that urine of humans and cats is rarely free of glucose, and glucose concentrations up to 1.4 mmol/L [25 mg/dL] and 1.5 mmol/L [26.7 mg/dL], respectively, are considered physiologic 12,17,18 Additionally the effects of possible influencing factors, including food intake, obesity, sex, and age were investigated.

| Study design
This study consisted of three parts, of which two were prospective (part 1 and 2), and one was retrospective (part 3). It included urine glucose measurements from healthy clinic-owned Beagle dogs (part 1), privately owned healthy normal-weight and obese dogs (part 2), and urine samples submitted to the University of Veterinary Medicine Vienna laboratory by a local practitioner (part 3). The study was approved by the institutional ethics and animal welfare committee in accordance with good scientific practice (GSP) guidelines and national legislation (10/12/97/2014 and ETK 09/03/2015).

| Impact of Feeding
Eight clinic-owned Beagles older than 1 year and housed in indooroutdoor runs were enrolled. There were six castrated and two intact male dogs from 2 years to 5 years (median 3 years) with a body condition score (BCS) of 5-7 of 9 (median 5). The body weights ranged from 14.4 to 21.6 kg (median 19.1 kg). Medical histories and physical examinations revealed no evidence of disease.
The experiment was performed over 4 days in familiar surroundings. On the first day, the dogs were accustomed to the handling procedures and were assigned to one of two groups (A and B) by simple block randomization using a shuffled deck of cards.

| Impact of obesity
Owners of 269 dogs aged 1 to 11 years, acquired via social media, public bulletins, or personal contact, who considered their dogs to be of normal weight (BCS 4/9) or obese (BCS 8 or 9/9), but otherwise healthy, were asked to fill out a questionnaire. To exclude dogs with diseases affecting urine glucose concentrations, the questionnaire included inquiries about the eating and drinking habits, orthopedic or endocrine problems, gastrointestinal and respiratory signs, recent stressful events, medications, and neoplasia. The return rate was 54% (n = 146), and 40 dogs were excluded. From the questionnaire, the reasons for exclusion were medications (n = 9), neoplasia, or a history of malignant tumors (n = 8), cardiorespiratory signs (n = 7), polyuria/polydipsia (n = 6), age (n = 5), recent stressful events (n = 3), skin alterations (n = 3), lethargy (n = 3), gastrointestinal signs (n = 2), liver disease (n = 1), bladder stones (n = 1), the BCS was too low (n = 1), and spinal problems (n = 1). Thirty-eight dogs left the study because owners were unable or unwilling to collect urine samples.

| Statistics
Statistical analyses were performed with the laboratory software package IBM SPSS Statistics 24. Normal distributions of data were tested with the Kolmogorov-Smirnov test. If data were not normally distributed, nonparametric tests were applied. The differences between groups (male vs female dogs and obese vs normal-weight dogs) were analyzed using the Mann-Whitney test. In the case of dependent variables (impact of feeding), the Wilcoxon signed-rank test was applied. Post-hoc Bonferroni-Holm corrections were used in part 1 of the study (impact of feeding) to account for multiple comparisons. To avoid gender bias in part 2 of the study, male and female distributions were compared using the chi-square test. Correlations between UG and other parameters (eg, age, weight, plasma glucose, urine protein, urine specific gravity) were tested using the nonparametric Spearman's rank correlation coefficient. After excluding outliers, defined as data higher than three interquartile ranges (3 IQR) above the third quartile (Q3), the 0.975-fractiles were calculated. Data are given as the median and range, and the level of significance was set at P < .05.

| Impact of feeding
All dogs ate the meals within 5 minutes, and the collection of adequate urine samples was possible at most sampling points. When fasted, urine production decreased, and urine collection was not possible at every sampling point. Eggs of Capillaria plica were detected in the urine sediment of four dogs. These dogs showed no signs of inflammation, for example, hematuria or active sediments.
Feeding did not affect UG concentrations or the UGCRs (Figure 1; Table 1). Glucose was detectable in all urine samples using wet chemistry analyses, but no dog was glucosuric using the Combur 9 dipstick test.
UG and the UGCR were comparable between lean and obese dogs ( Figure 2; Table 2). Low, but nonsignificant correlations were found between UG and BCS (r SP = .178, P = .242) and UG and weight (r SP = −.198, P = .191). UGCR correlated positively with BCS (r SP = .320, P = .032) and negatively with weight (r SP = −.405, P = .006). Glucose was detectable in all urine samples, but no dogs were glucosuric using the Combur 9 dipstick test.

| D ISCUSS I ON
Using a highly sensitive quantitative UG assay, we detected small amounts of glucose in 99.1% of 560 canine urine samples, which contradicts the paradigm, that urine is virtually glucose free provided the renal tubular maximum for glucose reabsorption is not exceeded. 1 Only five (2.2%) of the 227 urine samples submitted by a local small animal practitioner tested negative for glucose when analyzed by wet chemistry. Negative glucose measurements were not observed in healthy Beagle dogs irrespective of the feeding status, nor were they observed in obese and normalweight healthy privately owned dogs. Accordingly, the terms "normoglucuria" or "basal glucosuria" as established in humans 17,18,21 and suggested for cats, 12 seem appropriate for use in the canine species. Nevertheless, as in nondiabetic humans 17  As shown in eight healthy clinic-owned Beagle dogs, with multiple measurements over 5.5 hours, UG variability was small, and food intake had no significant short-term effects. The lack of UG variability and food effects was unexpected. Glucose reabsorption and consequently, glucose excretion is regulated mainly by sodium-glucose cotransporters 2 (SGLT 2) located at the apical membrane of renal proximal tubular cells. 22 These low affinity but high capacity glucose transporters are insulin sensitive and reabsorb about 90% of glucose together with insulin-independent basolateral GLUT2 transporters under normal conditions. 22 The elimination of insulin receptors expressed on renal tubular cells, 23 reduces SGLT 2 expression and increases UG excretion. 24 Human patients with increased insulin concentrations are more likely to have low UG concentrations independent of blood glucose concentrations. 25 As insulin increases postprandially, and peak blood glucose concentrations do not exceed the so-called "renal threshold" in healthy dogs, 26 a drop in postprandial UG concentrations was expected.
Although median UG concentrations dropped after meal intake, The whiskers indicate the range of values below and above the first (Q1) and third (Q3) quartile up to 1.5 times the interquartile range, respectively. Values lying more than 1.5 or 3 times below Q1 or above Q3 are outliers and depicted as dots or stars, respectively. The difference between the groups was not found to be significant (P = .421) [Color figure can be viewed at wileyonlinelibrary.com] the differences did not reach statistical significance.  29 In a study of obese diabetic fatty rats, UG concentrations increased 10 weeks before blood glucose levels exceeded renal thresholds, which was after the rats had received 12 weeks of a high-fat diet that nearly doubled their body masses. 30 As in cats, 12 no significant sex or castration status effects were observed. This was of interest as the expression of the SGLT-2s exhibits sex and species differences. The SGLT-2 protein shows higher expression in female rats than male rats. It is enhanced by estradiol and downregulated by androgens. Interestingly, the increased expression in female rats was not accompanied by higher SGLT-2-dependent glucose uptake on brush-border vesicles compared with male rats, suggesting a functional contribution of another glucose transporter system. In contrast to the rat kidneys, the SGLT-2 protein is dominant in male mice. 31 In a human study, including 261 healthy volunteers, no gender differences were found between males and females for up to 50 years, but women above this age had significantly lower mean urine glucose concentrations. 32 In a large Japanese study including participants with a mean (± SD) age of 63.5 (± 8.4) years, only 28.3% of the non-DM subjects with pathologic glucosuria were female. 29 In contrast to cats, 12 no correlations were found between the ages and UGs in the dogs of this study. Human studies that included healthy subjects 33 or type-2 diabetic patients 34 demonstrated rising renal glucose thresholds with age. Accordingly, higher blood glucose concentrations are needed in aged humans to override the renal capacity to reabsorb glucose. It is possible that our study was underpowered to show an age effect.
Although the data suggest breed differences, the number of dogs in each group was too small to allow reliable conclusions. Norwegian Elkhounds 1 or Basenjis, 3 two breeds with known predispositions for renal glucosuria were not included in the study population.
The determination of the UG concentrations does not incorporate urine flow rates, which are modulated by hydration status and renal reabsorption of free water. To allow for dilutional effects, we integrated urine creatinine measurements and calculated the UGCR.
Although this is already standard practice for other urine solutes such as proteins 35 and corticoids, 36 the authors found only one canine study where the UGCR was calculated to document glucosuria.
Five healthy Beagles were artificially made glucosuric by sequentially increasing constant rate glucose infusions. Significant increases in the UGCRs above baseline were observed at serum glucose concentrations of 10-11.1 mmol/L (180-200 mg/dL). 37 In this study, food intake, sex, and castration status did not affect the UGCR; however, the UGCR was positively correlated with age and BCS. Thus, relative age and BCS-dependent increases in glucose vs creatinine concentrations have to be considered when interpreting UGCRs.
In conclusion, our study results suggest that small amounts of glucose are continually present in canine urine and that glucose excretion is unaffected by the feeding status, obesity, age, or gender.
Prospective studies, including a larger group of healthy dogs, are encouraged to define cut-off values between physiologic and pathologic glucosuria.

D I SCLOS U R E
The authors have indicated that they have no affiliations or financial involvement with any organization or entity with a financial interest in, or in financial competition with, the subject matter or material discussed in the article.