Correlation between urine anion gap and urine ammonia‐creatinine ratio in healthy cats and cats with kidney disease

Abstract

Background

Ammonium excretion decreases as kidney function decreases in several species, including cats, and may have predictive or prognostic value in patients with chronic kidney disease (CKD). Urine ammonia measurement is not readily available in clinical practice, and urine anion gap (UAG) has been proposed as a surrogate test.

Objectives

Evaluate the correlation between urine ammonia‐to‐creatinine ratio (UACR) and UAG in healthy cats and those with CKD and determine if a significant difference exists between UAG of healthy cats and cats with CKD.

Animals

Urine samples collected from healthy client‐owned cats (n = 59) and those with stable CKD (n = 17).

Methods

Urine electrolyte concentrations were measured using a commercial chemistry analyzer and UAG was calculated as ([sodium] + [potassium]) − [chloride]. Urine ammonia and creatinine concentrations had been measured previously using commercially available enzymatic assays and used to calculate UACR. Spearman's rank correlation coefficient between UAG and UACR was calculated for both groups. The UAG values of healthy cats and cats with CKD were assessed using the Mann‐Whitney test (P < .05).

Results

The UAG was inversely correlated with UACR in healthy cats (P < .002, r ₀ = −0.40) but not in cats with CKD (P = .55; r ₀ = −0.15). A significant difference was found between UAG in healthy cats and those with CKD (P < .001).

Conclusions and Clinical Importance

The UAG calculation cannot be used as a substitute for UACR in cats. The clinical relevance of UAG differences between healthy cats and those with CKD remains unknown.

Abbreviations

Ca²⁺

calcium

CKD

chronic kidney disease

Cl⁻

chloride

HCO₃ ⁻

bicarbonate

K⁺

potassium

Mg²⁺

magnesium

Na⁺

sodium

PO₄ ³⁻

phosphate

SO₄ ²⁻

sulfate

UACR

urine ammonia to creatinine ratio

UAG

urine anion gap

1. INTRODUCTION

The kidneys play an important role in maintaining the body's acid‐base balance, which is critical for normal health. One of the kidney's key functions in acid‐base homeostasis involves reabsorbing filtered bicarbonate and generating new bicarbonate, which occurs through the process of net acid excretion. The 2 normal components of daily acid excretion are titratable acidity and ammonium excretion, which lead to the formation of new bicarbonate. Ammonium excretion is particularly important because it can increase considerably under conditions of metabolic acidosis as a compensatory response. ¹ , ²

Disorders of acid‐base homeostasis can result in a number of clinical problems such as electrolyte abnormalities, bone disorders, muscle weakness, and cardiac arrhythmias. In humans with progressive chronic kidney disease (CKD), a substantial proportion of individuals (up to 40%) develop metabolic acidosis, where inadequate capacity for ammonia excretion appears to be the driving force. ³ The gold standard in human medicine for evaluation of ammonia excretion is direct measurement of urinary ammonia, which then is reported as either urine ammonia‐to‐creatinine ratio (UACR) or daily excretion (if a 24‐hour urine result is known). However, 24‐hour urine collection is impractical in veterinary patients. Studies investigating 24‐hour urinary ammonia excretion in humans with CKD have shown that impaired ammonia excretion was associated with worse clinical outcomes, including more rapid progression of CKD and death. ⁴ , ⁵ , ⁶ , ⁷

Metabolic acidosis is also commonly recognized in cats with CKD and a previous study of acid‐base status in cats with CKD showed a tendency for decreased ammonium excretion in advanced stages of CKD, suggesting decreased capacity for cats with CKD to compensate for acidosis (Brown et al, ACVIM Forum 2023). ⁸ , ⁹ The UACR can be used to provide an estimate of ammonia eliminated in the urine over a 24‐hour period. A reference interval for UACR has been established for healthy cats (Brown et al, ACVIM Forum 2023).

Measurement of urine ammonium concentration is not readily available in veterinary practice settings. Because urinary ammonium may have prognostic value, more easily measured values or calculations such as urine anion gap (UAG) have been investigated to determine if they could be appropriate surrogates. ⁶ , ¹⁰ , ¹¹ , ¹² , ¹³ , ¹⁴

The UAG is calculated by subtracting measured anions from measured cations. One of the simplest and most commonly used calculations is (Na⁺ + K⁺) − Cl⁻ = UAG. These electrolytes all can be measured using commonly available commercial laboratory assays. Some of the most prevalent unmeasured anions in urine include sulfate (SO₄ ²⁻), phosphate (PO₄ ³⁻), and organic anions (eg, citrate). Some of the most prevalent unmeasured cations in urine are ammonium (NH₄ ⁺), calcium (Ca²⁺), and magnesium (Mg²⁺). ¹⁰ In humans, the simple UAG calculation described above has been shown to be a poor surrogate for urinary ammonium concentration in patients with CKD. ⁶ , ¹⁵ , ¹⁶ , ¹⁷ Measured and unmeasured anion and cation concentrations in feline urine may be significantly different from those of humans, and therefore, the correlation between UAG and urine ammonia in cats also may be different. ¹⁸ , ¹⁹ , ²⁰

Our primary aim was to evaluate the correlation, if any, between urine ammonium and UAG in healthy cats and cats with CKD and determine if UAG could be used as a surrogate marker for UACR in cats. A secondary objective was to determine if UAG differed between healthy cats and those with CKD.

2. MATERIALS AND METHODS

2.1. Study population

We utilized 2 sets of banked urine samples. The first set (healthy cat population) was collected from cats presented to the University of Florida (UF) primary care service for wellness examination. These cats were placed in the healthy category based on a lack of findings suggestive of clinically relevant illness based on the medical history, physical examination findings, serum biochemistry, and urinalysis. Cats were excluded from this category if they had azotemia or another disease likely to alter renal function (eg, hyperthyroidism), even if no clinical signs were reported. The second set (CKD population) was collected from cats with International Renal Interest Society (IRIS) stage 2 to 4 CKD presented to the UF nephrology/urology clinic. Cats were eligible for inclusion in the CKD group if they had a serum creatinine concentration >1.8 mg/dL or serum symmetric dimethylarginine (SDMA) concentration >18 μg/dL and lack of appropriate urine concentration (urine specific gravity <1.045) on at least 2 occasions when stable (<25% change between measurements) and well‐hydrated based on physical examination and serial body weight measurements. Cats from both groups were excluded if: (1) urine samples had macroscopic hematuria, ≥5 WBC/high power field or bacteriuria; (2) they had received glucocorticoids, diuretics, urinary acidifiers, alkalinizing agents, or angiotensin converting enzyme (ACE) inhibitors, or (3) insufficient urine sample was available (<100 μL). All cats (healthy and CKD) were fasted overnight before sample collection.

2.2. Sample collection

Urine samples were collected by cystocentesis and multiple aliquots were placed into separate cryotubes, which were stored at −80°C within 6 hours of collection. ²¹ For urine ammonia and creatinine measurements, single aliquots of each sample were thawed in batches and centrifuged at 2100g for 2 minutes (Thermo Scientific Sorvall Legend micro21R). ²¹ For measurements of sodium, potassium and chloride, separate aliquots of each sample were thawed at room temperature for 30 minutes and mixed for 1 to 3 seconds (Mortexer Vortex Mixer, 115 V). A minimum of 100 μL of urine from each sample was sterilely pipetted into 1 mL sterile conical tubes and delivered to the UF Small Animal Clinical Pathology Laboratory for analysis.

2.3. Data collection and analytical methods

Demographic data including age, breed, sex, and weight were recorded. The serum bicarbonate concentration was recorded and cats were classified as having metabolic acidosis if the concentration was ≤16 mEq/L. For the CKD cats, stage and substages (based on IRIS guidelines for serum creatinine concentrations, systolic blood pressure, and urine protein‐to‐creatinine ratio) were recorded, as well as whether or not they were fed a renal diet.

Urine ammonia and creatinine concentrations were measured using commercially available enzymatic assays (Ammonia Reagent Assay; Pointe Scientific, Canton, Michigan and Creatinine Assay Kit [ab204537]; Abcam, Cambridge, Massachusetts). The coefficient of variation (CV) and total observed error (TE_obs) for both assays have been determined previously. ²² The intra‐assay CV was 1.27%, the inter‐assay CV was 2.34%, and allowable total error (TEA) was 6.0% for ammonia. For creatinine, the intra‐assay CV was 1.27%, the inter‐assay CV was 2.54%, and the TEA was 3.5%.

Concentrations of Na⁺, K⁺, and Cl⁻ in urine were measured using an indirect ion‐selective electrode (ISE) with a commercially available chemistry analyzer (AU480, Beckman Coulter). The UAG was calculated using the formula (Na⁺ + K⁺) − Cl⁻ = UAG.

2.4. Statistical analysis

Data was compiled in an Excel (Microsoft Office, 2019) spreadsheet and analyzed using commercially available software (SPSS, IBM, Armonk, New York). Clinicopathologic data were tested for normality using Kolmogorov‐Smirnov test for the healthy population (n > 50) and Shapiro‐Wilk test for the CKD population (n < 50) and presented as mean (±SD) if normally distributed, or as the median (range) if non‐normally distributed. Demographic variables (age and weight) and clinicopathologic variables (urine Na⁺, K⁺, Cl⁻, UACR, and serum HCO₃ ⁻) were compared between healthy cats and those with CKD using an unpaired Student's test or Mann‐Whitney test depending on normality testing. A Chi‐squared test was used to compare proportions of sex in each group. Correlation between UAG and UACR variables was evaluated using Spearman's correlation test. Statistical significance was set to P < .05.

3. RESULTS

Fifty‐nine healthy cats, and 17 cats with CKD were considered for inclusion and none were excluded. The healthy cats included 32 male neutered, 26 female spayed, and 1 female intact cats. Mean age was 7.8 ± 3.9 years. Most were mixed breed including domestic short hair (n = 42), domestic medium hair (n = 2), and domestic long hair (n = 4). Purebred cats included Devon Rex (n = 3), Maine Coon (n = 2), Sphinx (n = 2), American Bobtail (n = 1), Bambino (n = 1), British Shorthair (n = 1), and Siamese (n = 1). Mean body weight was 5 ± 1.2 kg.

The cats with CKD included 11 male neutered and 6 female spayed cats. Mean age was 12 ± 4.9 years. Most were mixed breed, including domestic short hair (n = 11) and domestic long hair (n = 4). Purebred cats included a Persian (n = 1) and Ragdoll (n = 1). Mean body weight was 4.4 ± 1.1 kg. Twelve cats were classified as IRIS stage 2, 4 as IRIS Stage 3, and 1 as IRIS stage 4. Substaging information is included in Table S1. Renal diets were fed to 11 (65%) of these cats.

Demographic and clinicopathologic variable comparisons between healthy and CKD cats are included in Table 1. Healthy cats were younger and heavier than cats with CKD (P < .05). No significant difference was found between the proportions of sex in each group. Significantly lower median concentrations of Na⁺, K⁺, and Cl⁻ were found in cats with CKD compared with healthy cats (P < .001). Median UAG for cats with CKD was 40 (range, 20 to 156) which was significantly lower than the median UAG of 120 (range, −23 to 468) in healthy cats (P < .001; Figure 1). Urine anion gap was weakly negatively correlated with UACR in healthy cats (r ₀ = −0.40, P = .002; Figure 2), but no evidence of correlation was found in cats with CKD (r ₀ = −0.15, P = .55; Figure 2) or in the combined population (r ₀ = −0.11, P = .93; Figure 2). No difference in UAG was found in the cats with CKD that were fed a renal diet (median, 36.5; range, 21.5‐135.5) compared with those eating a non‐renal diet (median, 43.4; range, 20.1‐155.8; P > .05).

TABLE 1.

Demographic and clinicopathologic variable comparisons between healthy and CKD cats.

Variable	Healthy (n = 59)	CKD (n = 17)	P value
Age (years)	7.5 (±3.9)	12.0 (± 4.9)	<.001
Body Weight (kg)	5.2 (±1.2)	4.4 (± 1.1)	.02
Sex (n)	32 MN, 26 FS, 1 FI	11 MN, 6 FS	.44
UAG	120 (−23 to 468)	40 (20 to 156)	<.001
UACR	6.7 (3.0‐24.0)	3.1 (0.7‐5.3)	<.001
Urine Na+ (mEq/L)	178 (19‐371)	44 (10‐106)	<.001
Urine K+ (mEq/L)	159 (35‐315)	55 (27‐189)	<.001
Urine Cl− (mEq/L)	184 (27‐335)	53 (15‐131)	<.001
Serum creatinine concentration (mg/dL)	1.3 (0.5‐1.6)	2.5 (1.8‐5.0)	<.001
USG	1.052 (1.024‐1.060)	1.015 (1.009‐1.028)	<.001

Note: Data presented as mean ± SD for age and body weight and median (range) for serum bicarbonate and creatinine, USG, UAG, UACR, urine Na⁺, K⁺, and Cl⁻. Significant differences at P < .05 are bolded.

Abbreviations: Cl⁻, chloride; FI, female intact; FS, female spayed; K⁺, potassium; MN, male neutered; Na⁺, sodium; UACR, urine ammonia to creatinine ratio; UAG, urine anion gap.

FIGURE 1.

Box and whisker plot comparing UAG between the healthy and CKD population. The center line denotes the median value (50th percentile), the × represents the mean value, the box contains the 25th to 75th percentiles, the whiskers mark the 5th and 95th percentiles and circles beyond the upper and lower bounds are considered outlier cases. UAG is significantly lower in CKD cats compared with healthy cats (P < .01). Healthy; n = 59, CKD; n = 17.

FIGURE 2.

Correlation of UAG and UACR. Correlation between UAG and UACR (r ₀ = .11, P = .93) is shown for all cats (n = 76), including healthy cats (n = 59, represented by blue circles) and those with CKD (n = 17, represented by orange circles).

4. DISCUSSION

When UAG was calculated using only Na⁺, K⁺, and Cl⁻ only a weak inverse correlation with UACR was found in healthy cats, and no evidence of correlation was found in cats with CKD. Therefore, this simple calculation cannot be used as a surrogate for urine ammonia excretion in cats. Although some earlier studies in people suggested good correlation between the simple UAG calculation and urine ammonia, more recent studies have found correlations similar to what was identified in our study. ⁶ , ¹¹ , ¹³ , ¹⁵ , ²³ Some of these studies found that inclusion of additional anions such as sulfates and phosphates in the calculation provides better correlation with urine ammonia, but even with these additions, the UAG does not appear to predict clinical risks as well as direct measurement of urine ammonia concentrations. ⁶ , ¹⁰ , ¹⁵ Although including sulfates and phosphates may improve the correlation, it also requires measurement of these substances which, like measurement of ammonia, is not readily available in veterinary practice settings.

The UAG also can be affected by any discrepancy between gastrointestinal (GI) uptake and urinary excretion of the electrolytes measured in the equation. Therefore, certain conditions such as selective GI electrolyte loss from vomiting or diarrhea may independently impact UAG and not UACR. ²⁰ Measured and unmeasured anion and cation concentrations in feline urine are suspected to be considerably different from those of humans, particularly because of different contributions from dietary factors, urine concentration, and presence of different metabolites. ¹⁸ , ¹⁹ , ²⁰ Studies in people have shown that urine HCO₃ ⁻ concentrations can be considered negligible at urine pH <6.6, and therefore bicarbonate can be deleted from the UAG formula. It is unknown whether or not bicarbonate contributes substantially to UAG at urine pH >6.6. ¹⁰ However, it is possible that bicarbonate concentrations represent a more substantial contribution to UAG in cats than in people.

A significant difference was found between UAG in healthy cats and those with CKD. Interestingly, however, it was not in the direction that would be predicted if primarily a consequence of decreased ammonia excretion. With less ammonium excretion in CKD, a smaller contribution to the unmeasured cations would be expected and UAG should increase, but this group of healthy cats had significantly higher median UAG than did the cats with CKD. This finding was also consistent with more recent studies of people. ²⁰ Decades ago, the median UAG in healthy humans was reported to be 41 mEq/day ¹³ whereas the median UAG for patients with CKD was between 8 mEq/L and 26 mEq/day in different studies. ¹³ , ¹⁵ More recently reported values were significantly higher, with a UAG in healthy humans of >70 mEq/day ¹¹ and UAG for patients with CKD of 42 mEq/day. ⁶ This difference has been proposed as real change, likely associated with dietary differences. ¹¹

The finding of higher UAG in healthy patients suggests a large unmeasured anion contribution to the UAG in these patients. In patients with CKD, the previously unmeasured anion contribution either is substantially decreased, or there is a substantial increase in the presence of another unmeasured cation. The primary recognized unmeasured anions contributing to UAG are sulfate, phosphate, and organic anions. Studies in people have shown that phosphate excretion decreases with decreasing renal function, which may contribute to the lower UAG seen in the CKD patients, and this effect also may apply to cats with CKD. ²⁴ People consuming animal protein diets also have higher urinary excretion of phosphate and sulfate when compared with people eating vegetarian diets. ²⁵ Because cats with CKD often are started on a protein‐restricted diet (11 of 17 CKD cats in our study were on a renal diet), they may be excreting less urinary sulfate and phosphate (unmeasured anions), which could contribute to the overall lower urinary anion gap in the cats with CKD. Although a significant difference was not found between the UAG of cats with CKD fed a renal diet and those fed a non‐renal diet, there were small numbers of patients in these groups which may have impacted the power to detect a difference.

Urine sodium, potassium, and chloride concentrations all were significantly lower in the cats with CKD than in the healthy cats. Much of this difference may be attributed to loss of the kidney's ability to concentrate urine leading to dilution of urine electrolyte concentrations and a proportionally decreased UAG. However, dilution of electrolyte concentrations is not likely the primary factor contributing to low UAG because a post hoc analysis of correlation between UAG and USG found a low correlation coefficient, and some healthy cats with high urine specific gravities achieved lower UAGs than the cats with isosthenuria.

Our study had some limitations. A major limitation is the small number of cats with CKD, which prevented effective evaluation of potential differences in UAG or UACR in cats with different IRIS stages of CKD or subgroups based on factors such as diet or metabolic acidosis. Humans with renal tubular acidosis and respiratory alkalosis were shown to have decreased urinary ammonia excretion that strongly correlated with increased UAG. ²³ , ²⁶ In addition, humans with hyperchloremic metabolic acidosis caused by diarrhea, and humans and dogs with hyperchloremic metabolic acidosis caused by PO ammonium chloride administration had increased urinary ammonium excretion, which strongly correlated with decreased UAG. ²³ , ²⁷ A subset of cats with metabolic acidosis may have had a stronger correlation between UAG and UACR than what was observed in our study. However, the ability of cats to respond to experimentally induced metabolic acidosis when given ammonium chloride, has been questioned. ²⁸ Furthermore, in the studies of humans where ammonium chloride was given PO, at least some component of the low UAG may have been related to elimination of the additional Cl⁻ that was given. ²⁰ Another limitation is lack of direct pH and urine bicarbonate measurements. We evaluated the simplest equation used in people where there is a justifiable assumption that urine bicarbonate can be omitted from the UAG equation, but the same may not be true for cats, especially those with alkaline urine.

5. CONCLUSIONS

Based on the weak correlation, UAG calculated as (Na⁺ + K⁺) − Cl⁻, does not appear to be a useful surrogate for urine ammonia excretion in either healthy cats or cats with CKD. Some refinement of the UAG formula may provide better correlation, but likely would eliminate any practical value for using UAG instead of direct ammonia measurements. A significant difference between the UAG of healthy cats and cats with CKD was found, but further directed study would be necessary to help determine the underlying reason for and clinical relevance of this finding.

CONFLICT OF INTEREST DECLARATION

The authors declare no conflict of interest.

OFF‐LABEL ANTIMICROBIAL DECLARATION

Authors declare no off‐label use of antimicrobials.

INSTITUTIONAL ANIMAL CARE AND USE COMMITTEE (IACUC) OR OTHER APPROVAL DECLARATION

Informed consent was obtained from the owners or legal custodians of all animals described in this work (retrospective studies). No animals or people are identifiable within this publication; therefore, additional informed consent for publication was not required.

HUMAN ETHICS APPROVAL DECLARATION

The authors declare human ethics approval was not needed for this study.

Supporting information

Table S1. IRIS CKD stages and substages for CKD cat population.

Berman AR, Specht AJ, Castro RA, Cooke KL, Gilor S, Harris AN. Correlation between urine anion gap and urine ammonia‐creatinine ratio in healthy cats and cats with kidney disease. J Vet Intern Med. 2024;38(2):1068‐1073. doi: 10.1111/jvim.17002