Additional furosemide treatment beyond saline hydration for the prevention of contrast-induced nephropathy: a meta-analysis of randomized controlled trials

Abstract

Objectives: There have been conflicting results on the clinical utility of furosemide in preventing contrast-induced nephropathy (CIN). This study aimed to elucidate the effect of additional furosemide treatment beyond saline hydration on CIN post radiologic procedures by a meta-analysis of randomized controlled trials (RCTs). Methods: The Medline, EMBASE, and Cochrane databases were systematically searched. Two reviewers independently determined the eligibility of studies that randomly assigned patients undergoing radiologic procedure to receive additional peri-procedural furosemide injection or intravenous saline hydration alone. Combined results were presented as risk ratios (RR) with 95% confidence intervals (CI) by random-effect models. Results: We identified 5 RCTs including 1330 patients. Of them 659 received peri-procedural furosemide injection in addition to saline hydration, and 671 only received intravenous saline hydration (the control). Relatively small total number of dialysis events and heart failure events were reported in the included studies (n = 18 across 5 trials, n = 24 across 3 trials, respectively). Compared to the control, additional furosemide treatment did not significantly increase the incidence of CIN (RR = 1.18; 95% CI, 0.50-2.78; P = 0.71) and the risk of dialysis (RR = 1.03; 95% CI, 0.41-2.57; P = 0.95) post radiologic procedure. Furthermore, furosemide treatment appeared to decrease the occurrence of heart failure (RR = 0.35; 95% CI, 0.14-0.88; P = 0.02). Conclusions: Peri-procedural furosemide treatment in addition to saline hydration did not provide significantly prophylactic effect on CIN after radiologic procedure. Nevertheless, the process seemed likely to decrease the risk of heart failure post saline hydration.

Introduction

Contrast-induced nephropathy (CIN) is common in patients within two days after radiologic procedures with exposure to intravenous contrast media [1,2]. It can result in the increased need for acute dialysis and the extended hospital stays, and it is associated with substantial morbidity and mortality [3]. As a result, the preventative measures for CIN has attracted wide attention among cardiologists and radiologists [4]. Currently, prophylactic intravenous hydration with isotonic saline solution has been confirmed to be a protective strategy against CIN [5]. Of note, saline hydration is usually performed at a lower rate due to the fear of over-hydration and pulmonary edema, especially in these special patients with preexisting renal insufficiency or impaired left ventricular function [6,7]. Furosemide, a potent loop diuretic, has been combined with hydration therapy to increase urine output and prevent fluid overload [8,9]. More importantly, in addition to increasing urine flow, furosemide is a direct renal vasodilator and also increases tubular flow rates, results in greater contrast dilution within the renal tubules, decreases the metabolic workload of the kidney and subsequently decreases oxygen demand as well as prevents or ameliorates ischemic injury [10]. Most of these effects can target important components of the complex physiopathology of CIN. As thus, in theory, furosemide should be able to help to reduce the incidence of CIN after radiologic procedures. However, previous clinical studies have presented conflicting results on the benefit of additional peri-procedural furosemide treatment beyond saline hydration for prevention of CIN. A recent randomized controlled trial (RCT), the MYTHOS (Induced Diuresis With Matched Hydration Compared to Standard Hydration for Contrast Induced Nephropathy Prevention) study, suggested that compared with standard hydration furosemide-induced diuresis with matched saline hydration significantly lowered the risk of CIN in patients undergoing coronary procedures [10]. In contrast, the studies by Dussol B et al [8] and by Shemirani H et al [11] indicated that furosemide was associated with the increased risk for CIN after the exposure to radiocontrast agents. The unfavorable effect of furosemide was also found in two previous meta-analyses [3,12], which only included two small-scale furosemide-treatment studies. However, due to the limited study number and size, the previous meta-analyses did not provide relatively solid evidence for clinical practice. In light of the current conflicting evidence on clinical utility of furosemide for preventing CIN, we performed a meta-analysis of currently available data from RCTs to elucidate the true effect of peri-procedural furosemide strategy in addition to saline hydration for the prevention of CIN post radiologic procedures.

Methods

Study search and selection criteria

We conducted a systematic literature search of MEDLINE, EMBASE and Cochrane databases from 1970 through December 2013. Complex search strategies were formulated using the following keywords: renal failure, kidney failure, acute kidney injury, contrast, CIN, furosemide, saline, sodium chloride, and hydration. Additionally, we combined this strategy with a manual search of reference lists from all identified relevant publications. Studies were considered for inclusion in the meta-analysis if (1) they were RCTs; (2) they investigated the effect of additional peri-procedural furosemide treatment beyond saline hydration versus intravenous sodium chloride alone on prevention of CIN after contrast use; and (3) they reported any of the following clinical events: CIN, the subsequent need for dialysis, or heart failure for volume expansion. Studies comparing the effectiveness of different strategies for preventing CIN were excluded. We considered articles published in any language.

Data extraction and quality assessment

To identify the eligible studies, two investigators (G.G, Y.Z) reviewed all citations in duplicate and independently. A standardized form was used to extract the data, such as study characteristics, demographic characteristics, treatment strategies, and CIN definition from each study. Predefined clinical endpoints were also recorded. Any differences or disagreements were resolved through consensus. All eligible trials were assessed by the following quality criteria recommended by the Cochrane Collaboration: sequence generation of the allocation; concealment of allocation; blinding of participants, personnel, and outcome assessors; use of intention to treat analysis; description of withdrawals and dropouts. Additionally, a numerical score between 0 and 5 put forward by Jadad and colleagues [13], on basis of the following criteria: method of randomization described adequately; allocation concealment; blinding of participants, personnel, and outcome assessors; description of withdrawals and dropouts; use of intention to treat analysis, was assigned as a measure of study design and reporting quality with 0 being the weakest and 5 designated the strongest.

Data synthesis

Treatment effects were reported as risk ratio (RR) with 95% confidence intervals (CI). Overall estimates of effect were calculated with a random-effects model. For studies with no event of interest in a treatment group, 1.0 was added to all cells for continuity correction [14]. Statistical heterogeneity was measured using the I² statistic with a scale of 0% to 100% (> 75% represented very large between-study inconsistency) [15]. Sensitivity analyses were conducted to test the robustness of the effect by omitting each trial one at a time from analysis and subsequently computing overall estimates for the remaining studies. The potential publication bias was qualitatively assessed using funnel plot method. Results were considered statistically significant at P < 0.05. The pooled analyses were performed with the RevMan 5.2 software (The Cochrane Collaboration, Copenhagen, Denmark).

Results

Our initial search yielded 279 citations. We excluded 274 studies by screening title, abstract, and full-text, on the basis of our predefined criteria, leaving 5 RCTs [8,10,16-18] that met the inclusion criteria (Figure 1). A total of 1330 patients who received radiographic procedures involving contrast agents were included in the meta-analysis. Of them, 659 patients received intravenous furosemid injection in addition to peri-procedural saline hydration, and 671 only received intravenous sodium chloride infusion. The mean age of patients in the individual trials ranged from 58.5 to 73.5 years, and there was a high prevalence of male (66% to 78%) and a high percentage of hypertensive patients (71.9% to 83%; Table 1). Among the 5 trials included in the present meta-analysis, three did not demonstrate the preventive effect of additional furosemide administration on CIN [8,17,18], whereas the other 2 trials suggested the benefit of diuretic therapy [10,16]. Of them, 4 recruited patients with the use of contrast agents for coronary angiography or angioplasty [10,16-18]. Moreover, apart from one study [16], all of other 4 trials included patients with chronic renal deficiency. In terms of contrast type, two trials [8,10] used non-ionic, low-osmolality contrast agent, one only used ionic, low-osmolarity contrast [17], one used ionic, high-osmolality (32%), ionic, low-osmolality (33%), non-ionic, low-osmolality (35%) contrasts [18], and Gu et al. [16] did not specially report the contrast type. Contrast volume used during radiological procedures ranged from 100 to 181 milliliters. All of the included five trials reported a total of 180 CIN events (83 in furosemide group/97 in hydration group) and 18 dialysis events (9/9), and three of them totally reported 24 heart failure events (5/19). Assessments of methodological quality and Jadad scores were shown in Supplementary data (Table 2). The level of quality for each article was graded with a Jadad score of 1 to 5.

Figure 1.

Flowchart of identification of eligible trials.

Table 1.

Baseline characteristics of the enrolled randomized trials

First author, year	No. enrolled	Coronary angiography, %	Treatment strategy	Hydration volume, L (T/C)	Urine output, L (T/C)	Age	Male, %	Diabetes mellitus, %	Hypertension, %	Heart failure	Chronic renal deficiencies,%	Renal function	Contrast volume, ml	Contrast type	CIN definition
Dussol B, 2006	79/77	32	Furosemide IV, 3 mg/kg, just after procedure	ND	ND	65	67.9	28.8	82	15.5	100	CrC between 0.25 and 1 ml/s (15 and 60 ml/min)	119/115	Non-ionic, low-osmolality	SCr > 0.5 mg/dl, at 48 h post-procedure
Gu GQ, 2013	422/437	100	Furosemide IV, 20 mg 10 min after procedure	5.4/5.5	3.9/3.1	58.5	72.4	20.6	71.9	1.2	25.2	Not restricted	100/100	ND	SCr > 25% of baseline, or > 0.5 mg/dl, at 48 h post-procedure
Majumdar SR, 2009	46/46	100	Furosemide IV drip, 100 mg just before procedure	4.2/3.0	3.8/2.3	64.5	77.2	37	77.5	17	100	SCr < 1.7 mg/dl	152/126	Ionic, low-osmolarity	SCr > 25% of baseline, or > 0.5 mg/dl, at 48 h post-procedure
Marenzi G, 2012	87/83	100	Furosemide IV, 0.5 mg/kg 60 min before procedure	3.9/1.7	3.2/3.1	73.5	78	39.5	83	0	100	eGFR < 60 ml/min/1.73 m2	181/158	Non-ionic, low-osmolality	SCr > 25% of baseline, or > 0.5 mg/dl, at 72 h post-procedure
Solomon R, 1994	25/28	100	Furosemide IV, 80 mg 30 min before procedure	1.4/1.6	2.2/2.1	65	66	52.8	NA	26.4	100	SCr > 141.4 mol/l (> 1.6 mg/dl) or CrC < 1 ml/s (< 60 ml/min)	132/125	Ionic, high-osmolality (32%), ionic, low-osmolality (33%), non-ionic, low-osmolality (35%)	SCr > 0.5 mg/dl, at 48 h post-procedure

C, control; CIN, contrast-induced nephropathy; CrC, creatinine clearance; eGFR, estimated glomerular filtration rate; IV, intravenous injection; L, liters; NaCl, sodium chloride; ND, no data; SCr, serum creatinine; T, treatment.

Table 2.

Subgroup analysis for the incidence of CIN

Clinical factors	Incidence of CIN
	No. of studies	RR (95% CI)	p value
Contrast volume
Higher dose	2	0.71 [0.10, 5.07]	0.73
Lower dose	3	1.69 [0.43, 6.73]	0.46
Contrast type
Non-ionic, low-osmolality	2	0.86 [0.08, 9.42]	0.90
Other type	3	1.41 [0.50, 3.97]	0.52
Heart failure
Very low percentage (0-1.2%)	2	0.44 [0.21, 0.92]	0.03
Low percentage (15.5-26.4%)	3	2.15 [1.37, 3.37]	0.0008
Prior renal dysfunction
High percentage (100%)	4	1.48 [0.52, 4.22]	0.46
Low percentage (25.2%)	1	0.57 [0.38, 0.84]	0.005

CI, confidence interval; CIN, contrast-induced nephropathy; RR, risk ratio.

In the meta-analyses, we found that additional peri-procedural furosemide treatment did not have statistically significant impact on the incidence of CIN post radiologic procedures relative to peri-procedure saline hydration alone (RR = 1.18; 95% CI, 0.50-2.78; P = 0.71, I² = 85%; Figure 2). Moreover, additional furosemide treatment did not present marked influence on the need for dialysis treatment due to CIN in comparison with the saline hydration alone (RR = 1.03; 95% CI, 0.41-2.57; P = 0.95, I² = 0%; Figure 3). Nevertheless, furosemide treatment strategy seemed likely to reduce the occurrence of heart failure after contrast administration (RR = 0.35; 95% CI, 0.14-0.88; P = 0.02, I² = 0%; Figure 4). Furthermore, omission of each trial one at a time from the analysis in sensitivity analyses did not have any relevant influence on all of the overall results. When the studies by Gu et al. [16] and Marenzi et al. [10] removed from the pooled analyses of the incidence of CIN, the statistical heterogeneity in the clinical outcome became disappeared (I² = 0%). In addition, funnel plots were performed for the incidence of CIN and essential symmetry was found (see Supplementary data), suggesting that there was no significant publication bias existing among the included studies.

Figure 2.

Effect of furosemide treatment on the incidence of CIN. CI, confidence intervals.

Figure 3.

Effect of furosemide treatment on the incidence of dialysis for CIN. Abbreviations as in Figure 2.

Figure 4.

Effect of furosemide treatment on the occurrence of heart failure. Abbreviations as in Figure 2.

Discussions

The present study found that peri-procedural furosemide administration in addition to saline hydration had little significantly influence on the incidence of CIN and the risk of dialysis compared to saline hydration alone. In addition, the diuretic therapy might decrease the occurrence of heart failure after saline hydration post radiologic procedures.

Saline hydration remains the cornerstone of CIN prevention [17,19]. Furosemide administration may have some positive effects when associated with hydration. Use of furosemide was able to enhance contrast dilution in the renal tubule through increased urine flow, blocked tubular sodium reabsorption in the medulla, and reduced tubular workload and concomitant oxygen requirement [10]. Consequently, these beneficial pharmacological functions might be associated with a renal protective effect against CIN [20]. After pooling the individual studies, the present meta-analysis showed a neutral finding of additional furosemide treatment in the prevention of CIN and the subsequent need for dialysis. Nevertheless, still the benefit of furosemide failed to be simply denied. Specially, the study by Gu et al. [16], the largest-scale trial included in the meta-analysis, did demonstrate the significant benefit of furosemide in preventing CIN. Of note, unlike other studies, the study [16] applied lowest-dose furosemide (20 mg) and lowest-volume contrast agents (100 ml) in relative young patients (mean age of 58.5 years). Moreover, it only included low or very low percentage of patients with chronic kidney insufficiency (about 25%) as well as with cardiac dysfunction (1.2%). On basis of the findings, we presumed that the favorable protective effect of additional furosemide treatment on CIN might be easily achieved possibly in relative low-risk patients. On the other hand, the differences in the fluid administered and the fluid balance achieved also likely influenced the results of the individual studies. The study by Marenzi et al. with positive result on CIN enrolled in the meta-analysis, used furosemide and matched hydration, and the furosemide group had a net positive fluid balance (a 2-fold higher intravenous hydration rate and a slightly higher urine output obtained in furosemide-treated patients when compared with controls), whereas the hydration group had a net negative fluid balance [21]. Conversely, the study by Solomon et al [18] with an unbeneficial effect of furosemide treatment used furesemide and hypotonic saline hydration, and the two groups did not differ significantly in total urinary output. That is to say, furosemide did not induce volume depletion. Since furosemide was given 30 minutes before the use of contrast agents, it may have caused systemic or renal hemodynamic changes that exacerbated those produced by the contrast agent itself [18]. Additional furosemide infusion prevents fluid overload resulting from volume expansion. It potentially reduced the risk of heart failure [4]. Three [8,10,16] of the included five trials in the present meta-analysis, reporting the data on the occurrence of heart failure, did not show the significant difference in the occurrence of heart failure between additional diuretic therapy and saline hydration alone. However, the pooling result indicated a potential benefit of furosemide treatment in lowering the incidence of heart failure. The difference between the overall result and the individual results seemed likely to be attributed to the enlargement of population sample in the pooling analysis. Of note, only 5 heart failure events in furosemide group and 19 in hydration group were found in the meta-analysis. The relatively small total number of events across the 3 included studies might overestimate the precision of the pooled findings and we should be cautious in interpreting this estimate. Therefore, the larger-scale studies are required to further confirm the impact of furosemide on the incidence of heart failure after saline hydration.

Several limitations of this meta-analysis deserved comment. We did not consider the impacts of furosemide dose on clinical endpoints due to absence of relevant data, and as a result, we still could not confirm whether diuretic therapy existed dose-specific effects on the incidence of CIN. Moreover, most identified trials in the meta-analysis were small-scale and not double-blinded. Subgroup analyses were not performed because of the limited study number and population size. Nevertheless, appropriate meta-analytic techniques with random-effects models were used to pool the effect variables, and sensitivity analysis further confirmed reliability and stability of the overall results.

In conclusions, on basis of the available data from RCTs published previously, additional peri-procedural furosemide administration beyond saline hydration appeared to have little impact on the prevention of CIN after radiologic procedures, and it did not increase the subsequent need for dialysis. Additionally, additional furosemide treatment appeared to lower the risk of heart failure for fluid overload following saline hydration. As thus, furosemide administration if needed is feasible to patients with the exposure of contrast agents, and the treatment might be effective when restricted to some specific patients. Therefore, more clinical trials are needed to resolve the uncertainties concerning the relative effectiveness of furosemide treatment strategy (e.g. drug dose, therapy timing), and to identify the specific patients who might achieve benefit in preventing CIN from additional diuretic therapy.

Disclosure of conflict of interest

None.