Effect of Loop Diuretics on the Fractional Excretion of Urea in Decompensated Heart Failure

Abstract

Background:

Fractional excretion of urea (FEUrea) is often used to understand the etiology of acute kidney injury (AKI) in patients receiving diuretics. Although FEUrea demonstrates diagnostic superiority over fractional excretion of sodium (FENa), clinicians often assume FEUrea is not affected by diuretics.

Objective:

To assess the intravenous loop diuretic effect on FEUrea.

Methods:

We analyzed a prospective cohort (n=297) hospitalized with hypervolemic heart failure at Yale New Haven Hospital System. FENa and FEUrea were calculated at baseline and serially after diuretics. The change in FEUrea at peak diuresis was compared with the pre-diuretic baseline.

Results:

Mean baseline FEUrea was 35.2% ± 10.5% and increased by a mean 5.6% ± 10.5% following 80 mg (40–160 mg) of furosemide equivalents (P < .001). The magnitude of change in FEUrea was clinically important as the distribution of change in FEUrea was similar to the overall distribution of baseline FEUrea. Change in FEUrea was related to the diuretic response (r= 0.61, P < .001), with a larger FEUrea increase in diuretic responders (8.8%, interquartile range [IQR]: 1.8–16.9) than non-responders (1.2%, IQR: −3.2 to 5.5; P < .001). Diuretic administration reclassified 27% of patients between low and high FEUrea groups across a 35% threshold. Neither change in FEUrea nor percentage reclassified out of a low FEUrea category differed between patients with and without AKI (P > .63 for both).

Conclusions:

FEUrea is meaningfully affected by loop diuretics. The degree of change in FEUrea is highly variable between patients and commonly of a magnitude that could reclassify across categories of FEUrea.

Acute kidney injury (AKI) is common in hospitalized patients with heart failure. A common concurrent therapy and/or precipitant of AKI are loop diuretics.^1-3 An important factor determining appropriate therapy for AKI is understanding the underlying mechanism, with a focus on differentiating hemodynamic or pre-renal changes in creatinine from true renal injury such as acute tubular necrosis (ATN).^4-6 In hemodynamic disorders, renal tubular function is intact and renal sodium conservation often is maximal. However, with disorders such as ATN renal sodium handling is perturbed leading to a reduced ability to reabsorb sodium. This pathophysiologic difference underlies the use of the fractional excretion of sodium (FENa) to differentiate causes of AKI. However, in the setting of diuretic use, there can be an uncoupling of renal sodium handling from tubular function.⁷ As a result, the fractional excretion of urea (FEUrea) has emerged as an alternative diagnostic approach because urea transport does not occur directly via sodium transporters.^8-10 Notably, an FEUrea threshold of 35% is cited in the Kidney Disease Improving Global Outcomes (KDIGO) AKI guidelines and seminal nephrology textbooks for differentiating AKI etiologies during diuretic therapy, with the preponderance of the literature supporting the diagnostic superiority of FEUrea over FENa in the setting of diuretic therapy.^8,11-14

In clinical use, FEUrea is often assumed to be unaffected by diuretics, but to our knowledge this has never been directly assessed in hospitalized patients receiving diuretics.^7,15 Renal sodium and water homeostasis and urea handling are intimately linked. Thus, a reasonable hypothesis would follow that FEUrea could be significantly impacted by loop diuretics. An ideal study design to address this question would be to administer a high dose of loop diuretics to unselected patients with AKI of unclear etiology, to definitively ascertain the effect on FEUrea as a diagnostic tool in this population. However, this trial design is not feasible because it would require administration of loop diuretics to patients who may have pre-renal AKI or would introduce severe selection bias by exclusion of patients with equipoise to the AKI etiology. As such, our approach was to study the effect of loop diuretics on FEUrea in an unselected population of patients undergoing clinically indicated loop diuretic therapy in which AKI is a common complication; hospitalized patients with acute decompensated heart failure (ADHF).

Methods

Patient Population

Patients receiving intravenous (IV) loop diuretics at the 2 main hospitals in the Yale New Haven Hospital System enrolled into a prospective cohort study on loop diuretics (NCT02546583) were included in the analysis. Inclusion criteria included a diagnosis of hypervolemic ADHF and treatment with IV loop diuretics. Enrollment could occur at any point during the hospitalization as long as the patient remained hypervolemic and continued IV loop diuretic therapy (median 1.0 days from admission). Exclusion criteria included incontinence, inability to comply with timed urine collections, or use of renal replacement therapies. All patients received IV bumetanide or furosemide, with the dose determined by the treating cardiologist. No additional diuretic agents other than medications listed in Table 1 were administered in the study window. Per protocol, patients could not receive a dose of loop diuretic after midnight on the day prior to the study, with study day diuretic administration at 9:00 ±1 hour, which avoided a carryover effect from prior diuretic therapy on FEUrea.^16,17 All patients provided written informed consent and the study was approved by the Yale Institutional Review Board.

Table 1.

Patient Characteristics at the Time of Baseline FEUrea Assessment

Characteristic	Total Population (n = 297)	Baseline FEUrea ≤ 35% (n = 154)	Baseline FEUrea > 35% (n = 143)	P
Patient demographics
Age (y)	64 ±14	65 ± 13	63 ± 15	.399
Male	191 (64)	85 (55)	106 (74)	.001
Black race	88 (30)	45 (30)	43 (30)	.774
Body mass index (kg/m2)	34 ± 10	34 ± 9	34 ± 11	.899
Left ventricular ejection fraction (%)	39 ± 19	39 ± 18	39 ± 19	.985
Vital signs and physical examination
Systolic blood pressure (mmHg)	117 ± 19	115 ± 18	119 ± 19	.113
Heart rate (bpm)	80 ± 15	81 ± 15	80 ± 16	.807
Weight change from admission (kg)	−2.5 (−4.5 to −0.8)	−2.3 (−4.2 to −0.7)	−2.6 (−5 to −1)	.225
Comorbidities, n (%)
CAD	160 (54)	84 (55)	76 (54)	.912
Diabetes	162 (55)	94 (61)	68 (48)	.023
Hypertension	260 (88)	135 (88)	125 (87)	.923
Laboratory values
Serum sodium (mEq/L)	135 ± 5	136 ± 4	135 ± 5	.450
N-terminal pro–B-type natriuretic peptide (pg/mL)	3697 (1841–7770)	3814 (1896–8276)	3163 (1577–6753)	.310
Serum albumin (g/dL)	3.6 ± 0.4	3.6 ± 0.5	3.6 ± 0.5	.901
Urine microalbumin/creatinine ratio (mg/mg Cr)	0.4 (0.1–1.5)	0.3 (0.1–1.1)	0.5 (0.2–2.4)	.140
Blood urea nitrogen (mg/dL)	34 ± 19	37 ± 19	30 ± 18	.003
Serum creatinine (mg/dL)	1.35 (1.05 – 1.69)	1.38 (1.05 – 1.77)	1.29(1.04 – 1.60)	.190
eGFR (mL/min/1.73m2)	56 ± 23	53 ± 22	60 ± 24	.009
>90	10%	7%	14%	.078
60–89	31%	27%	36%
45–59	24%	26%	21%
30–44	23%	27%	18%
15–29	11%	12%	9%
<15	1%	1%	2%
Urine sodium (mmol/L)	59 ± 29	53 ± 27	66 ± 31	<.001
Urine urea (mmol/L)	246 ± 126	230 ± 114	263 ± 136	.021
Medical regimen
IV loop diuretic dose in FE (mg)	80(40–160)	80(40–160)	80(40–160)	.136
ACEI, ARB, or ARNI	139 (47)	66 (43)	73 (52)	.138
Thiazide	32 (11)	19 (12)	13 (9)	.378
Aldosterone antagonist	76 (26)	39 (25)	37 (26)	.857
Beta blocker	191 (65)	104 (68)	87 (61)	.260
Digoxin	19 (6)	14 (9)	5 (4)	.053

Data are presented as median (quartile 1–3), mean ± SD, or n (%) unless otherwise specified.

ACEI, angiotensin converting enzyme inhibitors; ARB, angiotensin receptor blockers; ARNI, angiotensin receptor-neprilysin inhibitor; CAD, coronary artery disease; CKD, chronic kidney disease; FE, furosemide equivalents.

Urine Collection and Assays

Patients were asked to completely empty their bladder before administration of the morning IV loop diuretic dose. The baseline urine sample was collected from this urine. Following the administration of the IV loop diuretic dose, a timed 6-hour urine collection with thorough supervision by study personnel was conducted and terminated with a forced void. Spot urine samples were collected at 1, 2, and 6-hours post-diuretic, and a spot sample was also taken from the cumulative volume of urine collected throughout the 6-hour period. We chose a 6-hour collection period because the natriuresis from IV bumetanide or furosemide is largely complete within this time period.^18,19

Urine electrolytes were measured via indirect ion-sensitive electrodes on the Randox Imola clinical chemistry analyzer (Randox Laboratories, Ireland, UK). If urine sodium was below the limit of detection, samples were diluted 1:1 with normal saline and reanalyzed with the final result calculated after adjustment for the dilution. Urea and creatinine were measured using Randox reagents on the Randox Clinical Chemistry analyzer Imola as per the manufacturer’s instructions (Randox Laboratories).

Calculations

Estimated glomerular filtration rate (eGFR) was calculated by the chronic kidney disease epidemiology collaboration equation.²⁰ FEUrea was calculated using the equation: (urine urea × plasma creatinine) / (plasma urea × urine creatinine) × 100%. FENa was calculated using the equation: (urine sodium × plasma creatinine) / (plasma sodium × urine creatinine) × 100%. All serum and plasma laboratory values were obtained before the beginning of the urine collection and analyzed as described Urine Collection and Assays. Equipotent conversions between loop diuretics were done as follows: 40 mg IV furosemide = 1 mg IV bumetanide.^21,22 Sodium-based diuretic efficiency was calculated as the 6-hour cumulative sodium output per doubling of IV loop diuretic dose.²³ Urine output-based diuretic efficiency was calculated as the 6-hour cumulative urine output per doubling of IV loop diuretic dose.²⁴

Study Definitions and Outcomes

A FEUrea threshold of 35% was utilized, where FEUrea ≤ 35% was considered low FEUrea and >35% was defined as high FEUrea.^8,11,13,25 When evaluating the change in outcomes from the pre-diuretic baseline, the 2-hour urine sample was utilized as the peak diuresis time point. If the patient was unable to produce urine at 2 hours, the 1-hour sample was used instead. Of the 297 patients studied, 241 were able to produce a spot urine sample at 2 hours and the remaining 56 utilized the 1-hour urine sample as the peak diuresis comparator (Supplementary Fig. S1). Patients with a peak diuresis FENa greater than the median peak diuresis FENa were categorized as diuretic responders, and patients with a peak diuresis FENa less than or equal to the median peak diuresis FENa were categorized as diuretic non-responders. We chose a urine sodium-based metric of diuretic response instead of urine volume as urine sodium has demonstrated superior prognostic value and is a recommended metric for measuring diuretic response.^23,26,27

The primary outcome was the magnitude, direction, and distribution of change in peak diuresis FEUrea compared with the pre-diuretic baseline sample. To evaluate the clinical ramifications of changes in FEUrea, we quantified the percentage of patients in whom the FEUrea high or low grouping would be reclassified by the peak diuresis FEUrea moving across the 35% threshold compared with the baseline FEUrea classification. Additional outcomes included the difference in FEUrea changes between diuretic responders and non-responders, changes in urine electrolytes, FEUrea, and FENa during the 6-hour natriuretic period, and identification of characteristics that correlated with change in FEUrea. To investigate whether FEUrea changes following diuretic therapy differ during AKI, we conducted a sensitivity analysis to evaluate the change in FEUrea between patients with AKI at the time of analysis to those without AKI. AKI was defined as an increase in serum creatinine ≥0.3 mg/dL from the outpatient baseline serum creatinine. Baseline serum creatinine was defined as the lowest serum creatinine in the 180 days prior to enrollment. Utilization of the first or lowest inpatient serum creatinine frequently misclassifies AKI with poor prognostic discrimination.^28,29 We used an outpatient serum creatinine close to hospitalization, in agreement with current expert opinion and current AKI studies.^30,31

Statistical Analysis

Data with a normal distribution are presented as mean ± standard deviations and data with a skewed distribution are shown as median (quartiles 1–3). Categorical values are presented as frequencies and percentages. The difference between two groups was tested for significance with t test for normally distributed variables, and Mann–Whitney U test for skewed variables. Chi-square test was used to compare categorical variables. Correlations are reported as Spearman’s correlation coefficients (r). Univariate logistic regression was used to show the odds of having the high/low grouping FEUrea reclassified (ie, to have discordant FEUrea group with pre-diuretic FEUrea compared with post-diuretic FEUrea). To capture the nonlinear association between FEUrea and high/low FEUrea group reclassification, both pre-diuretic and post-diuretic FEUrea were each modeled with a restricted cubic spline function with 4 knots and presented graphically. A two-tailed P value < .05 was considered statistically significant. Analyses were performed using SPSS v21.0 software (SPSS, Inc, Chicago, IL) and Stata SE v14.0 (Stata Corp, College Station, TX).

Results

A total of 297 patients completed the protocol and were included in the analysis. Baseline characteristics are presented in Table 1. Patients were hypervolemic by exam and received a median IV furosemide-equivalent dose of 80 mg (40–160 mg). The median serum creatinine was 1.35 (1.05–1.69) mg/dL with a mean eGFR of 56 ± 23 mL/min/1.73m². In response to the IV loop diuretic, patients produced a median 6-hour cumulative urine output (UOP) of 908 mL (622–1340). From baseline to peak diuresis, the median urine sodium concentration increased (53 [34.5–79.9] to 96.7 [76.8–114.1] mmol/L) and the median urine urea (228.7 [141.4–333.9] to 64.1 [48.9–95.0] mmol/L) and urine creatinine (86.6 [48.0–132.6] to 22.6 [14.5–34.5] mg/dL) concentrations decreased.

The baseline median FEUrea was 34.4% (28.0%–41.7%) with 52% (n = 154) exhibiting an FEUrea ≤ 35% and the FEUrea >35% in 48% (n = 143) of patients at baseline. No difference in heart failure (HF) characteristics or medical therapies existed between baseline FEUrea groups, but those with a baseline FEUrea >35% had a higher eGFR (P = .009), higher baseline urine sodium (P < .001), and lower blood urea nitrogen (P = .003).

Following loop diuretic administration, FENa and FEUrea both changed (Fig. 1). FENa increased from a pre-diuretic baseline of 0.6% (0.3–1.4) to 4.6% (2.2–7.1) in the peak diuresis sample. FEUrea changed from a pre-diuretic baseline of 35.2 ± 10.5% to 40.8 ± 13.6 % at peak diuresis, a 5.6% ± 10.5% change secondary to diuretic administration (Fig. 2A; P < .001). The change in FEUrea was bidirectional, with an absolute value change of 8.9 % ± 7.9% from baseline. Change in FENa demonstrated the strongest correlation (r = 0.61, P < .001) with change in FEUrea, whereas cumulative 6-hour UOP (r = 0.32, P < .001), measures of diuretic efficiency (r = 0.3, P < .001), and eGFR (r=0.17, P = .003) demonstrated weaker correlations (Supplementary Table S1). Other measures of kidney function and loop diuretic dose did not correlate with change in FEUrea (Supplementary Table S1). The change in FEUrea was consistent across baseline FEUrea values between 20% and 40% (Supplementary Fig. S1). Based upon the predefined FEUrea threshold of 35%, 27% (n = 80) of patients were reclassified by their peak diuresis FEUrea from their baseline FEUrea classification. Of the patients reclassified, 79% (n = 63) were reclassified from low to high FEUrea due to increasing the FEUrea across the threshold and the remaining 21% (n= 17) were reclassified from high to low FEUrea compared with baseline FEUrea. One of every 2 patients with FEUrea of 40% that happened to be obtained after a loop diuretic was administered would have had a FEUrea <35% if the parameter was obtained before diuretics were given (Fig. 2B). We explored the clinical utility of moving the FEUrea threshold to 50% and employing a FEUrea “gray zone” of 35% to 50%. Of those (n = 195) with a FEUrea >50% or < 35% at peak diuresis, 42% (n = 82) had a baseline FEUrea in the “gray zone” or opposite FEUrea category prior to diuretic administration, limiting clinical application (Table 2).

Fig. 1.

Change in FEUrea and FENa during diuresis. Left illustrates the median (interquartile range [IQR]) FEUrea (%) at baseline, peak diuresis, and then at 6-hours. Right illustrates the median (IQR) FENa (%) at baseline, peak diuresis, and then at 6-hours.

Fig. 2.

(A) FEUrea distribution at baseline and change at peak diuresis. (B) Odds of FEUrea group reclassification at baseline and peak diuresis. (C) FEUrea distribution at baseline and change at peak diuresis by diuretic response. (D) Odds of FEUrea group reclassification by diuretic response. (A) The distribution of FEUrea at baseline before diuretic therapy using a histogram with FEUrea (%) on the bottom x-axis and frequency of the baseline FEUrea on the y-axis. The change in FEUrea is superimposed as a box-whisker plot with the change at peak diuresis displayed on the top x-axis. FEUrea changed 5.6% bidirectionally with an 8.9% absolute value change. (B) An FEUrea of 30% at baseline had the highest odds (60%) of having the FEUrea group reclassified (discordant from baseline classification) at peak diuresis by crossing the FEUrea threshold of 35%. In patients where the post-diuretic FEUrea is 35%–45% (High FEUrea group), there is a ~50% chance that their FEUrea was <35% prior to loop diuretic therapy (low FEUrea group). (C) Diuretic responders experienced a 7-fold greater change in FEUrea at peak diuresis compared with non-responders, indicating that FEUrea is increasingly affected when the diuretic intent of loop diuretic therapy is achieved. (D) When the baseline FEUrea is 20% to 35%, the odds of FEUrea group reclassification (discordant from baseline classification) by the post-diuretic FEUrea exceed 60% among diuretic responders. The odds of reclassification at a baseline FEUrea near 30% was 40% among non-diuretic responders. DR, diuretic responder; NDR, non-diuretic responder.

Table 2.

Different FEUrea Thresholds for Reclassification

FEUrea at Baseline (%)	FEUrea at Peak Diuresis (%)
	< 35%	35–50%	>50%
<35	91 (31)	47 (16)	16 (5)
35–50	16 (5)	51 (17)	49 (16)
>50	1 (0.3)	4 (1)	22 (7)

All values are presented as n (%)

To further characterize the impact of diuretic response on FEUrea, the change in FEUrea was compared in diuretic responders versus non-responders. Diuretic responders experienced a greater change in FEUrea from baseline 8.8% (1.8–16.9) compared with diuretic non-responders 1.2% (−3.2 to 5.5; P < .001; Fig. 2C). In patients whose baseline FEUrea is 20% to 35%, the odds of reclassification by the post-diuretic FEUrea exceed 60% when the patient has a (desired) adequate diuretic response (Fig. 2D). Diuretic responders and non-responders were reclassified most frequently from low to high FEUrea (83% [n = 39] and 73% [n = 24] respectively). Diuretic non-responders were more likely to have a post-diuretic decrease in FEUrea from baseline. Table 3 compares baseline and post-diuretic characteristics between diuretic non-responders with an increase in FEUrea and diuretic non-responders with a decrease in FEUrea from baseline. Similar to the overall cohort, non-responders with an increased FEUrea had greater increases in FENa, whereas those with a decrease in FEUrea had minimal increases in FENa. Patients experiencing a decrease in FEUrea were more likely to have higher natriuretic peptide concentrations and hypertension, but no other baseline characteristic or diuretic parameters differed.

Table 3.

Directional Change in FEUrea in Diuretic Non-Responders

	Decrease in FEUrea at Peak Diuresis (n = 63)	Increase in FEUrea at Peak Diuresis (n = 84)	P
Baseline characteristic
Systolic blood pressure (mmHg)	119 (104–150)	124(110–135)	.615
Heart rate (bpm)	81 (73–98)	82 (75–98)	.871
Weight change from admission (kg)	−1.7 (−4.7 to −0.6)	−2 (−3.7 to −0.3)	.864
Diabetes	35 (56)	49 (58)	.736
Hypertension	61 (97)	69 (82)	.006
CKD	28 (44)	28 (33)	.170
Serum Sodium (mEq/L)	135 (132–139)	134 (132–139)	.665
N-terminal pro–B-type natriuretic peptide (pg/mL)	4540 (2154–10,235)	2966(1153 –6414)	.019
Blood Urea Nitrogen (mg/dL)	15 (8–20)	13 (8–17)	.622
Serum Creatinine (mg/dL)	1.29 (1.04–1.64)	1.25 (1.02–1.54)	.386
BUN/SCr ratio	3.4 (2.6–4.7)	3.7 (2.8–4.6)	.604
eGFR (mL/min/m2)	56 ± 24	61 ± 23	.231
IV loop diuretic dose in FE (mg)	80 (40–160)	80 (40–150)	.892
Percentage increase from chronic oral loop diuretic dose in FE (mg)	100 (50–188)	75 (50–150)	.412
Concurrent ACEI, ARB, or ARNI	32 (52)	36 (43)	.294
Concurrent thiazide	8 (13)	11 (13)	.943
Concurrent aldosterone antagonist	18 (29)	25 (30)	.924
2-Hour characteristic
FENa (%)	1.9 (0.9–3.1)	2.6 (1.4–3.8)	.023
Change in FENa from baseline (%)	0.8 (0.1–2.0)	1.9 (0.9–3.1)	<.001
Change in FEUrea from baseline (%)	−4.4 (−7.6 to −1.5)	5.0 (2.5–9.3)	<.001
UOP (mL)	150 (80–275)	185 (102–250)	.335
6-Hour Characteristic
Cumulative 6-h urine sodium output (mmol)	46.3 (20.7–86.2)	56.3 (35.1–88.5)	.118
Cumulative 6-h UOP (mL)	780 (444–1060)	807.5 (579–1039)	.484
UOP-based diuretic efficiency	312 (165–571)	417 (255 – 614)	.074
Sodium-based diuretic efficiency	25.6 (9.9–51.1)	32.1 (17.8–50.3)	.102

Data are presented as median (quartile 1–3), mean ± SD, or n (%) unless otherwise specified.

ACEI, angiotensin converting enzyme inhibitors; ARB, angiotensin receptor blockers; ARNI, angiotensin receptor-neprilysin inhibitor; BUN, blood urea nitrogen; CKD, chronic kidney disease; FE, furosemide equivalents; SCr, serum creatinine.

Of the 297 patients investigated, 212 (71%) had an outpatient baseline serum creatinine available a median 90 days (43–138) prior to enrollment for AKI evaluation (Supplementary Fig. S2). In this subcohort, 37% (n = 78) met criteria for AKI on the day of analysis. Patients experiencing AKI had a higher serum creatinine (1.65 [1.37–2.08] mg/dL) than those without AKI (1.24 [1.01–1.57] mg/dL; P < .001) but a similar baseline FEUrea (34.4 ± 9.8% vs 35.7 ± 10.6%; P = .39), respectively. Patients with AKI had an equivalent change in peak diuresis FEUrea (5.2 ± 9.6% vs 4.5 ± 11.6%; P = .64) and reclassification of AKI etiology by peak diuresis FEUrea (27% vs 29%; P = .73) than those without AKI respectively (Supplementary Fig. S3).

Discussion

Contrary to traditional wisdom, FEUrea is significantly influenced by loop diuretic administration.⁷ In this cohort of hospitalized patients with hypervolemic ADHF receiving intravenous loop diuretic therapy, FEUrea commonly and significantly increased above the pre-diuretic baseline. The pre-diuretic to post-diuretic change in FEUrea was of a magnitude that reclassified the FEUrea grouping across the 35% threshold in more than one-quarter of patients. Importantly, the direction and magnitude of FEUrea change correlated closely with diuretic response. In patients with a robust natriuretic response, FEUrea consistently increased from baseline with a magnitude 7-fold greater than poor diuretic responders. Our findings indicate that the timing of diuretic administration and diuretic response can significantly influence FEUrea and raises questions if the clinical utility of FEUrea could improve by incorporating timing of loop diuretics into its application/interpretation.

Our study is additive rather than contradictory to prior studies evaluating FEUrea in the setting of diuretic therapy. Previous literature investigated the diagnostic utility of FEUrea in a cohort of patients with AKI of unclear etiology and recent diuretic exposure. In the majority of these studies the independence of FEUrea was inferred based on its superiority to FENa. Investigators did not directly assess the effect of active diuretic exposure and diuresis on FEUrea.^8-10,13 Notably, several previous investigations included patients with diuretic exposure in the past 24–48 hours, and some reports did not specify the timing.^8-10,13 It is highly unlikely that a short acting IV loop diuretic given 24 hours prior to assessment of FEUrea would have a meaningful effect. As such, unlike the current study in which all patients were at peak diuresis when FEUrea was queried, it is unclear to what degree these prior reports were truly influenced by loop diuretic exposure.

Urea handling in the kidney is complex and remains incompletely understood.^12,32 Urea movement within the kidney involves passive reabsorption in the proximal tubule and inner medullary collecting ducts, urea recycling pathways in the renal medulla, and active reabsorption via facilitated urea transporters.¹² HF medications and the severity of decompensation may alter passive urea handling through changes in the renin-angiotensin-aldosterone system, glomerular filtration, and tubular flow rate. HF medications were stable over the 6-hour study period, limiting their impact on urea handling in a patient across time points. Likewise, HF medical therapies did not differ between patients with an increase or decrease in FEUrea (Supplementary Table S1). Although mild to moderate worsening kidney function during diuresis in AHF is often the result of pre-renal hemodynamic changes in both hypervolemic and hypovolemic states, the pre-renal mechanisms, and thus the diuretic-induced changes to FEUrea, may differ between volume states. The magnitude of FEUrea change correlated with diuretic response in our study, with a greater change in FEUrea occurring in patients with a more robust diuretic response. This correlation does not imply urea and sodium handling are directly related. Diuretics could alter urea handling via changes in renal blood flow, disrupting medullary osmotic gradients driving urea reabsorption, or other unknown mechanisms. This complex relationship is highlighted by the bidirectional change of FEUrea following diuretic therapy in our study. From a diuretic-induced reduction in plasma volume, one would anticipate FEUrea might slightly decrease from baseline. Yet, a majority of patients experience an increase in FEUrea during diuresis, and 20% exhibited a significant decrease in FEUrea. Further investigation to elucidate the underlying mechanism of urea handling during loop diuretic therapy is needed.

In our cohort, prior to diuretic exposure the median FEUrea was 34.4%, which is similar to that reported in other HF populations.³³ Employing a dichotomous approach to FEUrea interpretation with a threshold of 35% allows mild changes in FEUrea to alter the clinical determination of AKI etiology in populations where values close to the cutoff are commonly found. We found that with a post-diuretic FEUrea of 40%, ~ one-half of patients would have had a pre-diuretic FEUrea < 35%. Concerns about the diagnostic accuracy of the 35% threshold are confluent with previously published reports: in a single-center study of hospitalized patients who sustained AKI from various insults, an FEUrea threshold of 35% performed inferiorly to an FENa cutoff of 1% in diagnosing pre-renal injury.¹³ A subsequent multicenter study of critically ill patients with AKI had similar findings of poor diagnostic accuracy of FEUrea compared with FENa, both in the overall study population as well as the subgroup that received diuretics.³⁴ The current findings add to this literature demonstrating the susceptibility of both dichotomous FEUrea thresholds and a “gray zone” of 35% to 50% commonly lead to reclassification of patients into different categories following loop diuretic administration.

Our study has several limitations that warrant discussion. Although ~40% of our population met criteria for AKI, this was an ADHF population receiving loop diuretics, not a population undergoing AKI evaluation with concern for hypovolemia. We investigated the influence of loop diuretics on FEUrea but not the accuracy of FEUrea in determining AKI etiology. In addition, enrollment in our study could take place anytime during hospitalization, with the majority of patients receiving loop diuretics prior to study inclusion. We did ensure that no patients were given loop diuretics in the 9 hours preceding enrollment, most patients had substantially longer diuretic-free periods, but we recognize that diuretic therapy earlier in their hospital course may have affected tubular cells and altered urea handling thus influencing the “pre-diuretic” FEUrea.

In conclusion, contrary to prevailing wisdom, FEUrea is meaningfully affected by loop diuretic administration. The degree of change in FEUrea is highly variable between patients and commonly of a magnitude that could reclassify the etiology of AKI. Additional research is warranted to understand the effect from the timing of diuretic exposure on diagnostic performance of FEUrea.