Urine sodium concentration after intravenous furosemide in dogs with acute congestive heart failure and correlation with treatment efficacy

1 INTRODUCTION

Congestive heart failure (CHF) is characterized by increased sodium (Na) and congestion in extracellular spaces, such as the pulmonary interstitium or body cavities. Loop diuretics such as furosemide and torsemide block the Na-potassium (K)-2-chloride (Cl) cotransporter and reabsorption of these electrolytes in the thick ascending loop of Henle, which results in natriuresis, diuresis and decongestion.¹ Responsiveness to loop diuretics has important clinical implications, because presence of residual congestion is a strong predictor of poor outcomes in humans with acute CHF.^2-4 In dogs, assessment of diuretic responsiveness primarily relies on subjective or semi-quantitative findings such as resolution of dyspnea or improvement of radiographic signs.⁵ Quantitative measurements of diuretic responsiveness, such as weight change and urine volume provide information regarding fluid loss but not necessarily natriuresis, which is critical for efficient decongestion.^{6, 7}

In humans with acute CHF, spot urine sodium concentration (uNa) is a quantitative measure of diuretic responsiveness and associated with important clinical outcomes such as duration of hospitalization and long-term prognosis.^8-11 Spot uNa measured 2 hours after IV diuretics allows prompt identification of resistant patients who might benefit from diuretic intensification or additional treatments to achieve optimal decongestion.^{4, 12} Studies in humans have shown that low spot uNa is significantly associated with longer duration of hospitalization, less diuresis, worsening renal function, increased use of thiazides, rehospitalization for CHF, and mortality as compared to patients with higher uNa.¹³ Thus, low spot uNa commonly is regarded as a marker of poor diuretic responsiveness, which can be present in as many as 28% of humans with acute CHF.^{2, 4, 13-15}

In dogs with acute CHF, treatment typically involves hospitalization for supplemental O₂ and parenteral diuretics, among other treatments. Studies investigating the role of uNa in dogs with CHF are limited, and if and how uNa correlates with baseline clinicopathological characteristics or predicts successful decongestion in dogs with acute CHF is unknown.^{5, 16, 17} We aimed to examine uNa after IV furosemide in dogs with acute CHF so as to better understand the correlation between uNa and medical history and clinicopathologic variables. Furthermore, we aimed to correlate uNa with the cumulative frequency of treatment success, defined as successful discontinuation of supplemental O₂ therapy (DCSO₂). The primary study hypotheses were that uNa after IV furosemide would be correlated with the duration of treatment with supplemental O₂ (time_O2) and cumulative frequency of DCSO₂ over the duration of hospitalization in dogs with acute CHF.

2 MATERIALS AND METHODS

Records of dogs presenting to the Emergency Service of the Ryan Veterinary Hospital of the University of Pennsylvania were reviewed. Dogs with a diagnosis of acute CHF based on combinations of radiographs, point-of-care thoracic ultrasound examination, echocardiography, and response to treatment were eligible. The selection of which tests to perform to arrive at the diagnosis was at the clinician's discretion. Other inclusion criteria included hospital admission for treatment with supplemental O₂ either by O₂ cage or nasal O₂, consultation with or transfer of care to the section of cardiology, and uNa obtained between 30 and 540 minutes after IV furosemide. The IV furosemide dose administered immediately preceding measurement of uNa was termed the index diuretic administration. Criteria involving the allowable time from index administration to uNa were arbitrarily determined based on the rapid onset and peak action of IV furosemide that occur 30 to 60 minutes post-administration and a common practice in our hospital of decreasing frequency of administration to every 8 to 9 hours (ie, q8h) after the initial series of IV doses.^{18, 19} Thus, the maximum allowable duration of time between index dose and urine sampling was a function of the maximum duration of time between 2 consecutive furosemide administrations typically used in our hospital. Exclusion criteria included any of the following: administration of furosemide parenterally by constant rate infusion, change in diagnosis to conditions other than acute CHF during hospitalization, and a treatment plan that did not include supplemental O₂.

The following information was extracted from the medical record: signalment including age, sex, breed, and weight at presentation and at the time of DCSO₂, dose and frequency of any chronically administered PO diuretics, results of hematologic or serum biochemistry tests performed before the index furosemide administration, including PCV and blood urea nitrogen (BUN), serum creatinine, total solids (TS), serum Na (sNa), serum K (sK), and serum Cl (sCl) concentrations, as well as dose and time of index furosemide administration, total dosage of parenteral furosemide administration before the index administration, uNa, urine K concentration (uK), and the ratio of uNa to uK (uNa : K).

Duration of hospitalization is used in studies of humans as a measure of diuretic responsiveness and treatment efficacy, and typically refers to the time from hospital admission to discharge. In veterinary cardiology studies, a standard definition of duration of hospitalization does not exist. In our study, we chose not to use the definition used in studies of humans because it could be influenced by variables related to our hospital workflow, such as time spent in waiting or examination rooms or other factors not related to treatment efficacy, such as when the owner could most conveniently pick up the dog from the hospital. In our study, we defined duration of hospitalization as time_O2, which represents the time from the start of supplemental O₂ therapy to the time of DCSO₂, which was an indicator of treatment efficacy insofar as the dog's decongestion was sufficiently alleviated such that the dog could rest comfortably in room air. We tested the sensitivity of our definition by examining 2 additional treatment durations. The first was the duration of time from the generation of the dog's admission paperwork to DCSO₂ (time_admit), which accounted for treatments that might have been administered before the start of supplemental O₂. The second was the duration of time from drawing of blood for initial biochemical analysis to DCSO₂ (time_blood). The drawing of blood was used as an indication that the dog was being actively attended to by the clinical staff.

3 RESULTS

Records of 59 dogs with acute CHF and uNa measurements examined between June 2020 and April 2023 were retrieved. Eight dogs were excluded, including 5 dogs that never required supplemental O₂, 2 dogs that received constant rate infusions of furosemide, and 1 dog with a uNa measurement that was performed <30 minutes after the index furosemide administration. The remaining 51 dogs were analyzed and included 27/51 (53%) males and 24/51 (47%) females with a mean age of 10.1 ± 2.7 years and median body weight of 7.5 kg (IQR, 5.6-10.2 kg). Nineteen different breeds were represented including 17/51 (33%) dogs of mixed breed, 5/51 (10%) Cavalier King Charles Spaniels, and 4/51 (8%) Havanese. The remaining 16 breeds each represented 3/51 (6%) or fewer dogs. Degenerative mitral valve disease or dilated cardiomyopathy was the cause of CHF in 44/51 (86%) dogs and 4/51 (8%) dogs, respectively. Patent ductus arteriosus, subaortic stenosis, and arrhythmogenic right ventricular cardiomyopathy accounted for CHF in 1 dog each. Twenty-one of 51 (41%) dogs were receiving diuretics PO at home, including 17/21 (81%) dogs receiving furosemide and 4/21 (19%) dogs receiving torsemide. The median PO furosemide and torsemide dosages were 4.0 mg/kg/day (IQR, 3.8-5.5 mg/kg/day) and 0.41 mg/kg/day (IQR, 0.28-0.62 mg/kg/day), respectively. Serum biochemistry and hematology findings at time of presentation are shown in Table 1. During hospitalization, 39/51 (76%) dogs received furosemide parenterally before the index administration (median, 4.0 mg/kg; IQR, 2.1-6.0 mg/kg). The median index dose was 2.0 mg/kg (IQR, 2.0-2.1 mg/kg; range, 1.0-5.0 mg/kg). The median time from index administration to urine collection was 140 minutes (IQR, 90-210 minutes). The mean uNa was 86 ± 41 mmol/L (range, 9-182 mmol/L). The mean uK was 39 ± 21 mmol/L (n = 42; range, 11-83 mmol/L).

No significant difference in mean uNa was found between male and female dogs (male, 83 ± 43 mmol/L; female, 89 ± 39 mmol/L; P = .64). Neither age nor weight was significantly correlated with uNa (age, R = −0.13; P = .36; weight, R = 0.14; P = .32). The correlations between uNa and serum biochemistry and hematology results are shown in Figure 1. Significant fair negative correlations between uNa and BUN and uNa and PCV were found (Figure 2). A significant fair positive correlation between uNa and sK, and significant moderate positive correlation between uNa and sCl was found (Figure 2). A fair but insignificant positive correlation between uNa and sNa (R = 0.28; P = .05) was found. The mean uNa of dogs receiving PO diuretics at home was significantly lower than that of dogs not receiving diuretics at home (receiving home diuretics, 64 ± 37 mmol/L; not receiving home diuretics, 101 ± 37 mmol/L; P = .001; Figure 3), but no significant correlation between uNa and total daily PO furosemide dose (R = 0.29; P = .20) was found. Five dogs were euthanized before DCSO₂. The mean uNa of dogs that were euthanized was significantly lower compared to dogs achieving DCSO₂ and subsequent discharge from the hospital (euthanized, 46 ± 34 pmol/L, n = 5; DCSO₂ and discharge, 90 ± 40 pmol/L, n = 46; P = .022). Neither the index furosemide dose (rho = −0.21; P = .14) nor time from index administration to urine collection (rho = 0.15; P = .29) were significantly correlated with uNa. Urine Na decreased as dose of prior furosemide increased (rho = −0.40; P = .011). Results of the exploratory multivariable analysis indicated that uNa was significantly associated with 3 independent variables, including a positive correlation with sCl and negative correlations with PCV and PO diuretics at home (Table 2). The correlations between uNa and BUN, sK, PCV, and total dose of parenteral furosemide before the index administration were not significant once adjusted for other variables in the multivariable analysis. For every 1 mmol/L increase in sCl, uNa increased 3.3 mmol/L (95% CI, 1.8-4.9 mmol/L). For every 1% increase in PCV, uNa decreased by 2.8 mmol/L (95% CI, 1.0-4.6 mmol/L). The uNa of dogs receiving PO diuretics at home was 20.7 mmol/L (95% CI, 1.3-40.0 mmol/L) lower than dogs not receiving diuretics at home. These 3 variables accounted for 53% of the variability in uNa.

Mean time_O2 was 20.4 ± 11.6 hours. Discontinuation of supplemental O₂ was achieved in 46/51 (90%) dogs. Five of 51 (10%) dogs were euthanized while still receiving supplemental O₂ (mean, 27.0 hours; range, 14-52 hours). The time_O2 was significantly different between 2 study cohorts created on the basis of median uNa. Dogs with low uNa had significantly longer mean time_O2 than dogs with high uNa (uNa <87 mmol/L, 24.2 ± 2.6 hours vs uNa ≥87 mmol/L, 16.6 ± 1.7 hours; P = .02; Figure 3). Similar results involving both of the secondary hospitalization durations were found for mean time_admit (uNa <87 mmol/L, 26.0 ± 2.6 hours vs uNa ≥87 mmol/L, 19.8 ± 1.6 hours; P = .05) and mean time_blood (uNa <87 mmol/L, 24.9 ± 2.7 hours vs uNa ≥87 mmol/L, 17.2 ± 1.8 hours; P = .02).

The cumulative frequency of DCSO₂ was significantly higher in dogs with higher uNa (𝜒² = 7.96, P = .01). The cumulative frequency in dogs with uNa ≥87 mmol/L was 28%, 88%, 96% and 96% at 12, 24, 38, and 48 hours, respectively, whereas the cumulative frequency in dogs with uNa <87 mmol/L was 28%, 42% 73%, and 85%, at the same time points (Figure 4). Similar findings for both of the secondary hospitalization durations were found (time_admit, 𝜒² = 6.63, P = .01; time_blood, 𝜒²=9.15, P = .003). Results of univariable regression are shown in Table 3. Lower uNa and uNa : K and higher PCV and index furosemide dose were associated with lower probability of DCSO₂, but uNa was the only variable independently associated with DCSO₂ after multivariable regression. The model using uNa had better fit than a model using uNa : K (uNa AIC, 220 vs uNa : K AIC, 226). For every 10 mmol/L decrease in uNa, the cumulative frequency of DCSO₂ decreased by approximately 12% (SHR, 0.88; 95% CI, 0.83-0.95; P < .001). Stacked cumulative incidence curves display the probability of DCSO₂ and death over the duration of treatment in dogs with high vs low uNa (Figure 5).

The median weight change during hospitalization was −3.3% (IQR, −6.8 to −1.2%; n = 33). A fair and significant negative correlation between weight change and uNa was found (rho = −0.41; P = .02; Figure 6), but no significant effect of weight loss on DCSO₂ was found (weight loss ≥3.3%, 22.6 ± 12.0 hours; weight loss <3.3%, 22.6 ± 14.2 hours; P = .93). No significant difference in cumulative incidence of DCSO₂ between dogs with high weight loss compared to dogs with low weight loss was found (𝜒²=.28, P = .6).

4 DISCUSSION

Our main finding was that uNa after IV furosemide was associated with treatment duration, treatment efficacy, and cumulative frequency of DCSO₂ during hospitalization for life-threatening clinical signs in dogs with acute CHF. In our study, frequency of DCSO₂ and time_O2 were proxies for diuretic responsiveness, success of decongestion, and duration of hospitalization. Our findings agree with previous studies in human patients with acute CHF. In 1 study,¹¹ patients with uNa ≤60 mmol/L had significantly longer median hospital stays (7 days vs 5 days) and higher frequency of incomplete decongestion at discharge (67% vs 56%) than patients with higher Na. In another study,²⁵ low uNa was associated with significantly longer median duration of IV diuretic administration (8 days vs 4 days) and duration of hospitalization (11 days vs 9 days). In a third study,²⁶ uNa <50 mmol/L was significantly associated with longer median duration of IV diuretic treatment (7 days vs 5 days) and 3-fold higher risk of death during hospitalization. Guidelines in human medicine recommend routine spot uNa in patients with acute CHF², and our results suggest spot uNa in dogs with acute CHF also has value.

Another finding of our study is that weight loss, which is a proxy for urine volume and diuresis, was associated with uNa but not significantly associated with cumulative frequency of DCSO₂. Our results are similar to those of studies in humans that found weight loss and urine volume inconsistently are associated with treatment efficacy or post-discharge morbidity and mortality.^27-29 Traditionally, natriuresis and diuresis after diuretics are regarded as occurring concurrently, but in our study and in others,^{8, 30, 31} the strength of correlation between weight change or urine output and uNa differed widely, which emphasizes the importance of considering uN, and not just weight loss, as a metric of diuretic efficiency. The relative importance of uNa is not immediately intuitive, considering that the cause of the acute clinical signs is fluid retention in the pulmonary interstitium or body cavities. Thus, fluid loss might be expected to be highly predictive of efficacy and outcome, but redistribution of existing body fluids, rather than volume overload, likely plays an important role in both the development and resolution of congestion.³² Nearly half of human patients experience no weight gain in the month before development of congestion and hospitalization, and as many as a quarter of hospitalized patients experience either minimal or no weight loss during treatment.^{33, 34} Furthermore, weight loss and urine output do not consistently correlate with clinical signs of treatment efficacy such as dyspnea relief.^{29, 35, 36} Without natriuresis, fluid is able to redistribute from the intracellular to interstitial space and congestion persists despite diuresis and weight loss. The redistribution of body water causes intracellular dehydration and can slow resolution of clinical signs, increase risk of worsening renal function, and prolong hospitalization.³⁷ Optimal diuretic responsiveness involves large volumes of urine containing large amounts of solute wherein weight change and uNa are strongly correlated and treatment rapidly alleviates congestion. Our study and as others, show that the presumption of adequate natriuresis on the basis of fluid or weight loss is not always correct, especially in the case of high renal Na avidity, diuretic resistance, or excretion of solute-poor urine.^{2, 31, 38} Under these circumstances, metrics strictly associated with fluid loss are not necessarily indicative of efficient response or improvement in clinical signs. One study⁹ in human patients with acute CHF demonstrated that total Na excretion and spot uNa were strongly associated with decreased mortality regardless of urine volume. After adjusting for urine volume, patients in the lowest tercile of Na excretion were 6.24 times more likely to die than patients in the highest tercile, and for every 10 mmol/L decrease in uNa, risk of death increased by 11%. In contrast, urine volume was not significantly associated with mortality after adjusting for uNa.

Other important findings of our study are the correlations between clinicopathologic variables and uNa in response to receiving furosemide. One of the strongest was the negative correlation between uNa and chronic PO diuretics at home, regardless of daily dose. Our finding mirrors results in studies of humans,^{26, 39} as well as previous studies in dogs. One study⁵ in dogs with mitral valve disease found significantly lower uNa in dogs receiving chronic PO loop diuretics compared to dogs in the untreated subclinical stage. Median uNa from the 2 studies are comparable (ours, 58 mmol/L; previous study, 42-46 mmol/L) and both results fulfill criteria for poor diuretic responsiveness in humans (ie, uNa <50-70 mmol/L). In both studies, not only was low uNa common in dogs receiving PO furosemide, but it was independent of dose, which raises important questions about the prevalence and risk for diuretic resistance in dogs receiving any amount of furosemide and not just those receiving relatively high doses. The high frequency of low uNa and presumed less responsiveness in dogs receiving PO furosemide might favor an initial acute CHF treatment strategy using relatively high doses of furosemide and more frequent use of more potent loop diuretics or combinations of diuretics and vasodilatory treatments.

Other variables independently associated with uNa included serum Cl concentration and PCV, which likely reflect the importance of Cl on tubuloglomerular feedback and intravascular volume, respectively.^{5, 40} Some studies in humans found that the magnitude of hypochloremia was larger than hyponatremia, which might reflect the fact that the Na-K-2Cl cotransporter in the loop of Henle takes up 2 Cl for every 1 Na ion.⁴¹ Chloride is the main electrolyte involved in the homeostatic or tubuloglomerular feedback system in the kidney.⁴² The salt sensing role of the macula densa cells is dependent on Cl to effectively regulate expression of Na-K-2Cl cotransporters, secretion of renin, and maintenance of glomerular efferent arteriolar tone and glomerular filtration rate. Low Cl tends to increase the renal Na avidity and decrease uNa. Intravascular volume also plays a role in kidney function, diuretic responsiveness, and the tendency for the kidneys to retain Na and fluid. Depletion of intravascular volume with an accompanying increase in PCV and decrease in glomerular filtration rate would lower renal Na excretion and uNa.

The retrospective design of our study meant that data collection and treatment over the duration of hospitalization, including the time from index furosemide administration to uNa measurement, was not standardized. Urine Na concentration was measured at different times after IV furosemide, and some dogs had received IV furosemide before the index administration. The wide range of times and treatments likely would bias toward the null, yet significant correlations between spot uNa and time_O2 and DCSO₂ were detected. Freedom from strict collection and measurement times would facilitate applicability of uNa in clinical practice. In humans, it is recommended to measure uNa 2 hours after IV diuretic administration,² but uNa obtained over a wide range of times remains significantly associated with important outcomes.^{10, 11, 43} In 1 study,¹⁰ low uNa detected as long as 24 and 48 hours after starting treatment was associated with poor weight loss, residual congestion, and worsening renal function. In the same study, the association between 1-year mortality and low uNa was as strong at 48 hours after initiation of treatment as it was at 6 hours. Thus, data in humans suggest that low uNa detected over a range of different times during treatment has prognostic value, which would simplify use of uNa in clinical practice.

Our study had some important limitations. There is no widely recognized definition of treatment or hospitalization time in veterinary cardiology. In our study, the robustness of our findings was supported by results of the secondary hospitalization durations. We were not able to fully evaluate other potential indicators of diuretic responsiveness such as uNa : K, fractional excretion of Na, urine specific gravity, electrolyte-free water clearance, or change in serum and urine electrolyte concentrations over time. Our study did not include other measures of disease severity, such as radiographic or echocardiographic heart size or consideration of medications other than furosemide. We were unable to detect or separate out the potential effect of comorbidities, such a primary airway disease, on results. Our study population only included dogs for which clinicians chose to perform uNa measurement, which could lead to selection bias and the generalizability of our findings to all dogs with acute CHF is unknown. Our study involved a single center, and our treatment protocols and practice might differ from other sites in important ways. Finally, we are unable to assess what, if any, actions clinicians might have performed based on their knowledge of uNa that could have influenced results.

Our results give rise to important unanswered questions that merit further study, including a fuller assessment of the timing of urine collection and measurement in relation to diuretic administration, whether serial monitoring of uNa or other related markers, such as uNa : K, provides clinically valuable information, whether treatment modifications guided by uNa improve short- and long-term clinical outcomes,⁴⁴ the potential role of uNa as a surrogate endpoint in studies comparing different diuretic drugs or strategies, and, finally, the need for a better understanding of any associations between uNa and other potentially important variables, including serum and urine Cl concentration, diet, hydration status, dose and frequency of other medications, and use of nonloop diuretics. In the absence of urinary catheterization, uNa is measured in voided urine samples, which represents the cumulative effect of diuretic administration over time. How this needs to be interpreted differently than more instantaneous uNa measurements reported in humans requires further study. Our results help formulate hypotheses that can be tested in prospective studies. The potential of uNa as a relatively simple and quantitative predictor of diuretic responsiveness warrants further investigation.

In conclusion, we identified a wide range of uNa after IV furosemide administration in dogs with acute CHF. Important predictors of low uNa included history of chronic PO diuretic administration, low sCl, and high PCV. Urine Na was a significant predictor of treatment duration and cumulative frequency of successful DCSO₂ and was the only variable independent of other clinicopathologic variables, such as BUN, PCV, TS, and weight loss. In contrast to uNa, weight loss was not associated with treatment duration or frequency of DCSO₂, emphasizing the importance of considering uNa. Urine Na holds promise as a measure of diuretic efficiency that can lead to improved clinical outcomes in dogs with CHF.

CONFLICT OF INTEREST DECLARATION

Mark A. Oyama consults for and sits on advisory boards for CEVA Sante Animale, which markets loop diuretics, angiotensin converting enzyme inhibitors, and aldosterone antagonists. There is no direct relationship between the scope of this activity and this study. No other authors declare a 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

Authors declare no IACUC or other approval was needed.

HUMAN ETHICS APPROVAL DECLARATION

Authors declare human ethics approval was not needed for this study.