Absence of tolerance and toxicity to high-dose continuous intravenous furosemide in haemodynamically unstable infants after cardiac surgery

Abstract

What is already known about this subject

• Continous i.v. infusion of furosemide is superior to intermittent administrations, especially in haemodynamically unstable infants, because it results in a more controlled diuresis (although doses are generally chosen rather low).

What this study adds

• High-dose continuous furosemide infusion is an effective treatment for volume overload in haemodynamically unstable infants.

• Development of tolerance to furosemide was not observed despite high doses and prolonged exposure.

• Maximum serum furosemide concentrations remained well below the presumed toxic concentration.

Aim

To evaluate a high-dose continuous furosemide regimen in infants after cardiac surgery.

Methods

Fifteen haemodynamically unstable infants with volume overload admitted to a paediatric intensive care unit were treated with an aggressive furosemide regimen consisting of a loading bolus (1–2 mg kg⁻¹) followed by a continuous infusion at 0.2 mg kg⁻¹ h⁻¹ which was adjusted according to a target urine output of 4 ml kg⁻¹ h⁻¹. Frequent sampling for furosemide concentrations in blood and urine was done for 3 days with simultaneous assessment of sodium excretion and urine output.

Results

The mean furosemide dose was 0.22 (± 0.06), 0.25 (± 0.10) and 0.22 (± 0.11) mg kg⁻¹ h⁻¹ on the first, second and third day, respectively. Median urine production was 3.0 (0.6–5.3), 4.2 (1.7–6.6) and 3.9 (2.0–8.5) ml kg⁻¹ h⁻¹, respectively, on the first, second and third day of the study. The target urine production was reached at a median time of 24 (6–60) h and this was maintained during the study period. The regimen did not result in toxic serum concentrations and was haemodynamically well tolerated.

Conclusion

High-dose continuous furosemide infusion for 72 h in haemodynamically unstable infants after cardiac surgery appears to be a safe and effective treatment for volume overload. Development of tolerance against the effects of furosemide and ototoxic furosemide concentrations were not observed.

Introduction

Intravenous (i.v.) furosemide therapy is an accepted way to reduce volume overload in paediatric patients after cardiac surgery [1]. There is evidence that continuous i.v. furosemide is superior to intermittent furosemide therapy, because continuous administration results in a more controlled diuresis [2–4]. However, the currently used continuous i.v. furosemide regimens are still largely empirical, especially in patients with varying renal function. Current practice is to start with a relatively low continuous i.v. furosemide dose of 0.1 mg kg⁻¹ h⁻¹ followed by higher doses up to 0.4 mg kg⁻¹ h⁻¹ as insufficient diuresis remains. As furosemide acts at the luminal site of the renal tubules, the driving force for furosemide to reach its site of action will be low in patients with a relatively low glomerular filtration rate. This may be the case in infants after cardiopulmonary bypass (CPB) surgery who are haemodynamically unstable and in whom transient renal insufficiency frequently occurs. Indeed, we have previously suggested that it may be more rational to start with a higher dose (0.2 mg kg⁻¹ h⁻¹) and adapt the dose guided by the urine output [5]. We have also reported the preliminary results of a pharmacokinetic–pharmacodynamic model to support this suggestion [6]. However, it has been shown that after repeated bolus administrations of furosemide, a diminished diuretic effect occurs [7, 8]. In addition, it has also been suggested that tolerance to the diuretic effect occurs during continuous infusion [9]. This may have an impact on our suggestion that a continuous i.v. infusion with a relatively high dose may be superior to a relatively low-dose regimen. Therefore we explored whether or not tolerance towards the diuretic effect of furosemide develops in haemodynamically unstable infants after cardiac surgery with CPB who are in need of diuretics.

Materials and methods

The study was performed at the Paediatric Intensive Care Unit (PICU) of Leiden University Medical Centre (LUMC). The study protocol was approved by the Committee on Medical Ethics of LUMC and conducted according to the principles of the Declaration of Helsinki. In addition, the trial was performed with the regulations on paediatric research set forward in Dutch law and in accordance with the recommendations for paediatric research of the Dutch Society of Paediatrics. Parental written informed consent was obtained for all patients.

Patients

Consecutive patients <1 year old who were admitted to the PICU of LUMC after CPB surgery were eligible if it was likely that they would need prolonged continuous i.v. furosemide because of volume overload and/or relatively low urine output (<4 ml kg⁻¹ h⁻¹), despite initial treatment with a bolus i.v. furosemide. All infants were haemodynamically unstable defined as at least 2/4 points for cardiovascular organ failure assessed with the modified Sequential Organ Failure Assessment Score [10]. The infants required inotropic support and this was quantified using the standard vasopressor score [11].

The infants were monitored with an arterial and a central venous line. During the postoperative course maintenance fluid was kept at 60 ml kg⁻¹ per 24 h. Volume expanders were administered when necessary to maintain or achieve adequate circulating volume and serum potassium levels were kept between 3.8 and 4.3 mmol l⁻¹, if necessary by supplementation. The observation period for this study was 3 days after the start of the continuous infusion.

The study protocol followed the routine clinical care for the patients as closely as possible and all drugs necessary for the treatment of the patients were allowed.

Furosemide regimen

When during the postoperative course the patients developed volume overload and/or insufficient urine output, first an i.v. furosemide bolus of 1 mg kg⁻¹ was administered. When this was ineffective, a continuous furosemide infusion of 0.2 mg kg⁻¹ h⁻¹ was started. This infusion was preceded by a loading dose of furosemide, which dose depended on renal function. Patients with normal renal function received 1 mg kg⁻¹ and patients with acute renal failure (ARF) received 2 mg kg⁻¹. ARF was defined as doubling of serum creatinine compared with preoperative serum creatinine or a serum creatinine concentration ≥75 µmol l⁻¹ in patients <8 weeks old [12]. The aim was to reach and maintain a urine output of 4 ml kg⁻¹ h⁻¹. It was therefore allowed to change the rate of the continuous infusion with steps of 0.1 mg kg⁻¹ h⁻¹. Adaptation of infusion rate was allowed at 12-h intervals and was based upon the urine production over the preceding 6 h. If the urine production was <2 ml kg⁻¹ h⁻¹ the rate of infusion was increased and if urine production was >6 ml kg⁻¹ h⁻¹ the infusion rate was decreased.

Sampling and assays

For blood sampling, it was taken into consideration that the total volume of blood taken for study purposes should not exceed 3% of the circulating volume. Blood and urine samples (using bags providing protection from ultraviolet light) were collected every 6 h for measurements of sodium, creatinine and furosemide. Serum sodium and creatinine concentrations were measured using a photometric method on an automatic analyser (Hitachi 747-100; Roche Diagnostics, Almere, the Netherlands). Furosemide concentrations were measured using a validated high-performance liquid chromatography method routinely applied at the laboratory of Clinical Pharmacy and Toxicology of LUMC. For determination in serum the coefficient of variation of the assay at 1 µg ml⁻¹ was 2%, and the reproducibility of the slope was 8.9%. For the analysis of furosemide concentration in urine, the samples were first deglucuronidated. The coefficient of variation of the assay in urine was 3.4% at 10 µg ml⁻¹, and the reproducibility of the slope was 7.2%.

Data analysis

Data showing a skewed distribution are given as median and range, whereas the normally distributed haemodynamic parameters are presented as mean and SD. The outcome evaluation included the median urine production over each 24-h time interval, the time at which the target urine production was reached and the deviation from the target urine production. The time to attain the target urine production was defined as the time point at which urine production was at least 4 ml kg⁻¹ h⁻¹ for two consecutive assessments. Deviation from the target urine production was defined as the absolute amount of urine either below or above target urine production.

The relationship of the time course of the serum furosemide concentrations, urinary furosemide excretion and urine production were displayed graphically to explore the possible development of tolerance. Blood pressure (BP) values and heart rate were summarized over time. The values obtained before the start of the furosemide infusion were compared with those obtained at the end of the experiment using paired Student's t-tests.

Results

General

Eighteen patients likely to require at least a 3-day treatment of continuous i.v. furosemide were included in the study. In three patients continuous i.v. furosemide was discontinued after 36, 30 and 60 h, because it was no longer indicated on clinical grounds. Thus, 15 patients completed the full 3-day course of furosemide infusion and these patients' data are reported (Table 1).

Table 1.

Patient characteristics

Patient	Diagnosis	Gender	Age (weeks)	Weight (kg)	CPB time (min)	Clamp time (min)	T-hypo (min)	T-low (°C)
1	IAA + VSD	F	1	3.3	122	60	107	19.5
2	PAIVS	M	0.5	3.5	33	19	21	34.6
3	DORV	F	12	4	121	91	127	26
4	MAPCA	F	6	3.2	60	29	93	29.4
5	DORV	M	23	5.8	112	69	118	27.4
6	TOF	F	12	4.9	102	76	111	30.2
7	TOF	F	30	5.6	148	68	180	28
8	TGA	M	1	3.7	108	68	104	25
9	AVSD	M	26	6.1	33	0	20	35.7
10	VSD	M	10	3	130	101	121	27
11	VSD	F	4	3.5	166	112	156	27.7
12	VSD	M	22	5.5	97	54	64	30.2
13	DORV	F	35	6.2	125	68	125	24.7
14	VSD	F	12	4.5	62	39	52	30.6
15	TOF + PAPVC	F	10	3.8	272	107	188	24.2

IAA, Interrupted aortic arch; VSD, ventricular septal defect; PAIVS: Pulmonary atresia with intact ventricular septum; DORV, double outlet right ventricle; MAPCA, major aortopulmonary collateral; TOF, tetralogy of Fallot; TGA, transposition of the great arteries; AVSD, atrioventricular septal defect; PAPVC, partial anomalous pulmonary venous connection; CPB time, time on cardiopulmonary bypass; Clamp time, aortic cross clamp time; T-hypo, hypothermia time defined as the period during which the body temperature of patient <36°C; T-low, lowest temperature reached during cardiopulmonary bypass.

The median (range) age of the infants was 12 (0.5–35) weeks and the weight was 4.0 (3.0–6.2) kg. All patients underwent major cardiac surgery with CPB. The median CPB time was 112 (33–272) min. Hypothermia was applied during surgery for a median period of 111 (20–188) min with a lowest body temperature of 27.7 (19.5–35.7)°C. At the start of the study, all patients were mechanically ventilated and haemodynamically unstable, requiring substantial inotropic support as shown by a mean (SD) vasopressor score of 23 (15).

There was clear volume overload in the patients, as shown by the mean elevated central venous pressure (CVP) of 15 (range 7–20) cm H₂O (Table 2).

Table 2.

Haemodynamic status and inotropic therapy at the start of the study

Patient	CVP (cm H2O)	SBP (mmHg)	DBP (mmHg)	HR (bpm)	Vasopressor score	Dobu (µg kg−1 min−1)	Dopa (µg kg−1 min−1)	Adr (µg kg−1 min−1)	Nor-Adr (µg kg−1 min−1)	Enoximone	NO (ppm)
1	8	75	58	151	10	5	5	0	0	4	14
2	8	65	30	145	25	10	10	0	0.05	0	0
3	18	70	40	195	23	5	12	0	0.06	3	0
4	10	55	35	215	25	10	5	0.1	0	5	0
5	20	80	55	170	25	15	10	0	0	0	0
6	13	75	45	145	18	10	8	0	0	0	0
7	18	65	35	125	30	10	10	0	0.1	0	0
8	7	65	45	160	11	5	6	0	0	0	0
9	15	70	40	154	17	5	12	0	0	0	10
10	17	88	42	124	42	10	10	0.14	0.08	2	14
11	15	80	50	170	30	10	10	0	0.1	2	0
12	19	92	45	150	7	5	2	0	0	0	0
13	13	75	40	182	14	5	3	0	0.06	0	0
14	8	98	55	160	5	5	0	0	0	0	0
15	15	72	52	201	66	0	10	0	0.56	4	22

CVP, Central venous pressure; S/DBP, systolic/diastolic blood pressure; HR, heart rate in beats per minute; Dobu, dobutamine; Dopa, dopamine; (Nor-)Adr, (nor-)adrenalin; NO, nitric oxide in parts per million.

Disconnection from mechanical ventilation occurred after a median time of 131 (70–215) h and discharge from the PICU was after a median time of 166 (86–257) h after surgery.

Aminoglycosides were administered to 10 patients. Routine therapeutic drug monitoring showed the concentrations of these drugs to be in the therapeutic range. No other drugs with a potential for interaction with furosemide kinetics (e.g. nonsteroidal anti-inflammatory drugs) or other drugs with a nephrotoxic potential were given. At the start of the study nine patients were diagnosed with ARF (as defined above). At the end of the 3-day observation period, ARF was still present in four infants, but in none of the infants had renal function deteriorated and all patients were discharged without evidence of renal insufficiency.

The mean (SD) total amount of blood taken for study purposes was 2.2% (0.5%) of the circulating volume.

Furosemide regimen

The mean (SD) total dose of furosemide boluses administered to the patients before the start of the continuous furosemide infusion was 2.94 (± 1.08) mg kg⁻¹. Continuous i.v. furosemide was started at a median time 25 (14–34) h postoperatively. The mean continuous furosemide dose was 0.22 (± 0.06), 0.25 (± 0.10) and 0.22 (± 0.11) mg kg⁻¹ h⁻¹ on the first, second and third day, respectively. There was no need to change the dose over the entire observation period for four patients. Dose adaptation, all increases, was made in four infants on day 1. On both day 2 and day 3, an increase in dose was needed in two patients, whereas the dose was decreased in four patients. Thus, there was a low need to adapt the doses within each day.

Furosemide kinetics

The mean furosemide concentrations over time are shown in Figure 1. Before the start of continuous furosemide administration the median serum furosemide concentration was 1.6 (0.2–3.4) µg ml⁻¹ and this increased to 5.2 (1–22.6), 3.9 (0.4–28.5) and 2.9 (0–23.2) µg ml⁻¹ on the first, second and third day, respectively. Of note, the maximal furosemide concentration observed was 28.5 µg ml⁻¹. Urinary furosemide excretion (Figure 1), which was relatively low at the first measurement at 6 h after initiation of the continuous infusion, increased rapidly and substantially in the next 18 h and remained stable thereafter. The median urinary furosemide excretion was 0.12 (0.04–0.22), 0.16 (0.1–0.27) and 0.15 (0.03–0.40) mg kg⁻¹ h⁻¹ for the three subsequent study days.

Figure 1.

Average (± SD) graphs of serum furosemide (µg ml⁻¹), furosemide excretion rate (mg kg⁻¹ h⁻¹), sodium excretion (mmol kg⁻¹ h⁻¹) and urine output (ml kg⁻¹ h⁻¹). In the two top panels, the boxes indicate the average furosemide dose with the height of the boxes proportional to the dose, with the first box indicating an infusion rate of 0.2 mg kg⁻¹ h⁻¹. In the bottom panel the target urine output (4 ml kg⁻¹ h⁻¹) is indicated by the dashed line

Furosemide dynamics—urine output and sodium excretion

The mean urinary sodium excretion and the urine production in each 6-h time interval over the entire observation period are given in Figure 1. Median sodium excretion was 2.5 (0.3–11.0), 7.1 (1.1–15.2) and 6.3 (1.6–17.5) mmol kg⁻¹ per 24 h over the first, second and third study days.

The median urine output before the start of the continuous infusion was 1.7 (0.2–7.6) ml kg⁻¹ h⁻¹. Median urine production over the consecutive study days was 3.0 (0.6–5.3), 4.2 (1.7–6.6) and 3.9 (2.0–8.5) ml kg⁻¹ h⁻¹ on days 1, 2 and 3. The target urine production was reached after a median time of 24 (6–60) h. The median deviation from the target urine production was 1.0 (0.5–3.4), 0.9 (0.1–2.6) and 1.1 (0.0–4.5) ml kg⁻¹ h⁻¹ for the three consecutive study days.

There was a strong linear relationship between sodium excretion and urine production, with correlation coefficients ranging from 0.66 to 0.99 (range of P-values 0.02–0.0003).

The relationship with the time course of the average group values of the serum furosemide concentrations, urinary furosemide and sodium excretion and the resulting urine production (Figure 2) indicates that there was no development of tolerance to the furosemide effect. With increasing urinary furosemide excretion, higher values of sodium excretion and urine output were observed, with no indication of tolerance. The data (illustrated in both graphs 1 and 2) also indicate that, after some time, the urinary furosemide concentration remained stable despite decreasing serum concentrations, indicating that the renal function of the infants improved. This is also shown by the serum creatinine concentrations, which decreased during the observation period from a median 95 (36–167) µmol l⁻¹ prior to the start of the infusion to 82 (41–188), 64 (42–192) and 61 (42–154) µmol l⁻¹ on, respectively, the first, second and third day.

Figure 2.

The relationship with the time course of the average group values of the serum furosemide concentrations, urinary furosemide and sodium excretion and the resulting urine production. The first observation is indicated with the closed symbols and the subsequent observations are connected with the lines

Haemodynamic and metabolic effects

The fluid balance was negative for all three study days, albeit with substantial variability. The median values were −31 (−270/+305), −139 (−272/+128), −37 (−790/+199) ml per 24 h for day 1, 2 and 3, respectively. Apart from the maintenance fluid, 60 ml kg⁻¹ per 24 h, volume expanders were needed in five, four and two patients on day 1, 2 and 3, respectively. The amount of volume expanders used varied between 10 and 80 ml per 24 h. The patients tolerated the forced diuresis well; BP increased slightly over the observation period (systolic BP +9.7 mmHg, P = 0.03; and diastolic BP +2.1 mmHg, P = 0.34), whereas the CVP (−3.5 cm H₂O; P = 0.007) and heart rate (−28.2 bpm; p = 0.007) decreased over time (Figure 3).

Figure 3.

Average (± SD) graphs of systolic ▴ (SBP), diastolic ▿ (DBP) blood pressure, central venous pressure ▪ (CVP) and heart rate ○ (HR)

Metabolic alkalosis, defined as pH > 7.45 and bicarbonate >29 mmol l⁻¹, was not observed during the entire study period.

Discussion

This study suggests that continuous high-dose i.v. furosemide is a well-tolerated, safe and effective means of reducing volume overload in haemodynamically unstable infants after CPB surgery. We noticed that in our patients the urinary excretion of furosemide increased and serum creatinine concentrations gradually decreased over time. We infer that these changes are an indication of improving renal function. This seems also to be reflected in the need for lower doses of furosemide towards the end of the 3-day observation period, while the urinary output remained stable. Thus, our observations do not support the suggestion that tolerance of the furosemide effect may develop with prolonged diuretic exposure as suggested by Eades et al.[9]. On the contrary, our data show that the diuretic effects, expressed as sodium excretion or as urine output, increased shortly after initiation of the therapy until 24 h and remained at least stable or even increased over the subsequent 48 h. The reason why either repeated bolus or continuous administration of furosemide results in tolerance for the diuretic is unclear. It has been suggested that interference with the autonomic nervous system, the renin–angiotensin–aldosterone (RAA) system [13] or atrial natriuretic peptide [14, 15] may play a role. However, the role of all these hormones has been shown to be minimal, if present at all [13, 15–17]. Therefore it has been suggested that the tolerance to furosemide can be induced through different but complementary homeostatic mechanisms in the kidney [16, 18]. Whatever the mechanism underlies the development of tolerance, it is clear that dehydration plays a major role. This provides a possible explanation why tolerance was not observed in our patient population, because they were volume overloaded and certainly not dehydrated at any time during the continuous furosemide infusion.

Our study supports observations that continuous i.v. furosemide results in a controlled diuresis [2–4]. We further observed that high-dose continuous i.v. furosemide therapy for 3 days was effective in achieving a negative fluid balance, but it was not associated with cardiovascular instability. On the contrary, BP increased while heart rate and CVP decreased over time. This is also supported by the remarkably low need for volume expanders.

Commonly occurring adverse events after high-dose furosemide were not noted in our study population. The regimen was associated with improvement of the transient ARF that often occurs in infants after cardiac surgery with CPB. Thus, it is unlikely that the high furosemide dose was associated with renal toxicity. In addition, we found that the relatively high furosemide doses did not result in metabolic alkalosis in our patient population. However, it has to be noted that our patients were mechanically ventilated, which may also have prevented the development of furosemide-induced metabolic alkalosis if it had occurred. Finally, despite the high furosemide doses, serum concentrations >50 µg ml⁻¹, which are generally considered toxic [19], were not observed. In our study, therapeutic drug monitoring for aminoglycosides was routinely performed and there was no indication that, at any time, toxic concentrations of these drugs were reached.

Although this may be an indication that screening for (oto)toxicity may be unnecessary in this population, caution is warranted when high-dose furosemide is coadministered with other ototoxic drugs such as aminoglycosides.

In summary, there are no indications that tolerance develops towards the diuretic effect of furosemide in haemodynamically unstable infants with volume overload after cardiac surgery with CPB, who are treated with a relatively aggressive diuretic regimen with furosemide. This strategy results in a negative fluid balance, while cardiovascular stability is not compromised, ototoxic concentrations are not observed and metabolic alkalosis does not occur.

The limitations of this study are that the study population was relatively small and that a study period of 72 h was employed. However, in clinical practice an indication for continuous i.v. furosemide rarely exceeds 72 h.

The current data suggest that the employed furosemide regimen can be used safely in haemodynamically unstable infants after cardiac surgery. However, it has to be confirmed in a randomized, controlled trial that this approach with a high starting furosemide dose is superior to a regimen employing a low starting furosemide dose.