The added value of hypertonic saline solution to furosemide monotherapy in patients with acute decompensated heart failure: A meta‐analysis and trial sequential analysis

Abstract We assessed the effects of hypertonic saline solution (HSS) plus furosemide versus furosemide alone in patients with acute decompensated heart failure (ADHF). We searched four electronic databases for randomized controlled trials (RCTs) until June 30, 2022. The quality of evidence (QoE) was assessed using the GRADE approach. All meta‐analyses were performed using a random‐effects model. A trial sequential analysis (TSA) was also conducted for intermediate and biomarker outcomes. Ten RCTs involving 3013 patients were included. HSS plus furosemide significantly reduced the length of hospital stay (mean difference [MD]: −3.60 days; 95% confidence interval [CI]: −4.56 to −2.64; QoE: moderate), weight (MD: −2.34 kg; 95% CI: −3.15 to −1.53; QoE: moderate), serum creatinine (MD: −0.41 mg/dL; 95% CI: −0.49 to −0.33; QoE: low), and type‐B natriuretic peptide (MD: −124.26 pg/mL; 95% CI: −207.97 to −40.54; QoE: low) compared to furosemide alone. HSS plus furosemide significantly increased urine output (MD: 528.57 mL/24 h; 95% CI: 431.90 to 625.23; QoE: moderate), serum Na+ (MD: 6.80 mmol/L; 95% CI: 4.92 to 8.69; QoE: low), and urine Na+ (MD: 54.85 mmol/24 h; 95% CI: 46.31 to 63.38; QoE: moderate) compared to furosemide alone. TSA confirmed the benefit of HSS plus furosemide. Due to the heterogeneity in mortality and heart failure readmission, meta‐analysis was not performed. Our study shows that HSS plus furosemide, compared to furosemide alone, improved surrogated outcomes in ADHF patients with low or intermediate QoE. Adequately powered RCTs are still needed to assess the benefit on heart failure readmission and mortality.


| INTRODUCTION
Currently, acute decompensated heart failure (ADHF) represents a high burden of morbidity and mortality worldwide. 1 Treatment of fluid overload is a major issue in the management of ADHF patients, with loop diuretics (e.g., furosemide) being the cornerstone therapy. 2,3 In some patients, usual doses of loop diuretics are not enough to relieve symptoms of congestion, and up to half of the patients leave the hospital with persistent fluid overload which is associated with rehospitalizations and higher mortality. 4 This insufficient response, known as diuretic resistance, has been shown to contribute to worsening heart failure during hospitalization, prolonged lengths of stay, and increased mortality. [5][6][7] Adding other diuretics, such as thiazides or metolazone, acting in different sites in the nephron is a common practice trying to address diuretic resistance. 1 However, this strategy requires careful monitoring of electrolytes, especially sodium and potassium, and renal function. Thus, use of hypertonic saline solution (HSS) in combination with furosemide has emerged as a novel therapeutic approach. 8 The addition of HSS to furosemide is based on its property to recall free water contained in the interstitial spaces, increasing the intravascular compartment during diuretic therapy. 9 Subsequently, HSS can prevent the decline in effective arterial circulating volume and the consequent possible decrease in renal blood flow, being the main reason to be considered in scenarios like refractory decompensated heart failure. 8 Therefore, we performed a systematic review and meta-analysis to assess the effects of HSS plus furosemide versus furosemide alone in ADHF patients.

| METHODS
This systematic review was reported following the 2020 Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRIS-MA) statement. 10 Ethical approval was not required because this study will retrieve and synthesize data from already published studies.

| Search strategy
We searched in the following electronic databases from inception to January 29, 2021, with an update on June 30, 2022: PubMed, Embase, Scopus, and Web of Science. The complete search strategy is available in Supporting Information: Table S1. There were no restrictions on language or publication date. We also performed a hand search of reference list of all included studies and relevant review articles to identify other potentially eligible studies.

| Eligibility criteria
The inclusion criteria were as follows: (i) randomized controlled trials (RCTs) involving adult patients (≥18 years old) with ADHF, (ii) RCTs evaluating any dose and duration of HSS plus furosemide as the intervention group, (iii) any dose and duration of furosemide as the control group, and (iv) RCTs that report at least one evaluated outcome at any length of follow-up. Observational studies, case reports, case series, systematic reviews, preprints, conference abstracts, and editorials were excluded.

| Selection of studies
We downloaded all articles from electronic search to EndNote X8 and duplicate records were removed. All unique articles were uploaded to Rayyan (https://rayyan.qcri.org/) for the study selection process. Titles and abstracts were independently screened by two review authors (Carlos Diaz-Arocutipa and Jack Denegri-Galvan) to identify relevant studies. Furthermore, the same review authors (Carlos Diaz-Arocutipa and Jack Denegri-Galvan) independently examined the full-text of selected studies and registered reasons for the exclusion. Any disagreement on title/abstract and full-text selection was resolved by consensus.

| Outcomes
The outcomes were classified as follows: clinical outcomes (all-cause mortality, cardiovascular mortality, and heart failure readmission), intermediate outcomes (length of hospital stay, urine output, and weight), and biomarker outcomes (serum Na + , urine Na + , serum creatinine, and type-B natriuretic peptide [BNP]). We used the studyreported definitions for all outcomes.

| Data extraction
The information from each selected study was independently extracted by two review authors (Carlos Diaz-Arocutipa and Jack Denegri-Galvan) using a standardized data extraction form in an Excel spreadsheet that was previously piloted. Any disagreement was resolved by consensus. If additional data was needed, we contacted the corresponding author through email. The following data were extracted: first author name, publication year, country, study design, sample size, population, age, sex, comorbidities, left ventricular ejection fraction (LVEF), intervention group, comparator group, and clinical, intermediate, and biomarker outcomes.

| Risk of bias assessment
Two review authors (Carlos Diaz-Arocutipa and Jack Denegri-Galvan) independently assessed the risk of bias in each study using the Cochrane risk of bias (RoB) tool 2.0. 11 Any disagreement was resolved by a third author (Adrian V. Hernandez). The RoB 2.0 tool evaluates five domains: randomization process, deviations from intended interventions, missing outcome data, measurement of the outcome, and selection of the reported result. Overall, each RCT was judged as having a low, some concerns, or a high risk of bias.

| Assessment of the quality of evidence
We used the grading of recommendations, assessment, development and evaluation (GRADE) approach to evaluate the quality of evidence for each outcome. 12 The GRADE methodology examines the following five categories: risk of bias, consistency, indirectness, imprecision, and reporting bias. Each RCT started as high-quality evidence and will be downgraded based on the criteria described above. The quality of evidence was categorized as high, moderate, low, or very low. We generated the summary of findings (SoF) table using the GRADEpro software.

| Statistical analyses
All meta-analyses were conducted using the inverse-variance random-effects model. Treatment effects were expressed as relative risk (RR) with their 95% confidence interval (CI) for dichotomous outcomes and mean difference (MD) with their 95% CI for continuous outcomes. 13 Only final values for each group were compared for urine output, serum Na + , urine Na + , serum creatinine, and BNP. For weight, the difference between final and baseline values was compared. Only studies that evaluated outcomes with similar follow-up times and cointerventions were pooled. Heterogeneity was evaluated using the chi-squared test (threshold p < .10) and the I 2 statistic, with values of I 2 > 60% corresponding to substantial Furthermore, we conducted a trial sequential analysis (TSA) to evaluate the random errors due to multiple testing and sparse data, and to calculate the required information size. 14 Our calculation, defined a priori, was based on the autogenerated empirical data according to the data input for continuous outcomes, two-sided type I error of 5%, and statistical power of 80%. We also calculated the TSA-adjusted CI for all outcomes. This analysis was performed using the TSA software version beta 0.9.5.10 (Copenhagen Trial Unit).

| Study selection
Our electronic search yielded 756 articles. After the removal of 333 duplicates, 423 articles underwent title/abstract screening, of which 28 articles were selected for the full-text screening. Eighteen articles were excluded by the following reasons: conference abstract (n = 10), other study design (n = 5), editorial (n = 1), duplicate data (n = 1), and other intervention (n = 1). Finally, 10 articles were included ( Figure 1). 15-24

| Trial characteristics
The main characteristics of the 10 RCTs (n = 3013) are summarized in Table 1. The mean age was 70 years and 60% were men across trials.
A total of six out of 10 studies were conducted in Italy and the rest in subjects, 64%). 21 In six studies, the patients were blinded to the treatment. While in five studies, it was the treating physicians. Only in one study 15 were the investigators blinded and three studies were open label. In half of the studies, only hospitalized patients with refractory ADHF were included, while in the rest, unselected cases were included. In seven studies, the duration of the HSS treatment was <6 days. Mean LVEF ranged from 24% to 56%, with mean LVEF < 40% in eight studies. Comorbidities were reported in eight studies, although information was mostly limited. The most common etiology of heart failure was ischemic cardiomyopathy with a prevalence ranging from 42% to 62%. Mean baseline serum Na + ranged from 135 to 139 mmol/L. The daily dose of administered furosemide in both groups varied between 500 to 2000 mg/day, which corresponded to the trials conducted in Italy, while in non-Italian trials, varied between 40 and 200 mg/day. Information on the timing of outcome assessment for each study is available in Supporting Information: Table S2.

| Risk of bias assessment
Overall, five RCTs were judged as some concerns as the risk of bias (Supporting Information: Figure S1). Five RCTs showed some concerns in the deviations from the intended interventions, two RCTs showed concerns in the measurement of the outcome, and two RCTs showed some concerns in the selection of the reported result.
The other five RCTs were judged as low risk of bias. DIAZ-AROCUTIPA ET AL. | 855

| Clinical outcomes
Only four RCTs evaluated the effect of HSS plus furosemide compared to furosemide alone on all-cause mortality, cardiovascular mortality, and readmissions for heart failure. 16,20,21,23 The timing of clinical outcomes assessment ranged from 30 days to 57 months across studies (Supporting Information:   Figure 2). The quality of evidence was moderate ( Table 2). TSA-adjusted CI: −3.25 to −1.39; I 2 = 24%) compared to furosemide alone (Figure 2). The quality of evidence was moderate (Table 2)

| Trial sequential analysis
For all outcomes, the trial sequential monitoring boundaries were crossed by the cumulative z-curves and the TSA-adjusted CIs showed a significant beneficial effect of HSS plus furosemide over furosemide alone (Supporting Information: Figures S2 and S3). The diversityadjusted relative information size was only reached for the following outcomes: length of hospital stay, weight, urine output, serum Na + , and urine Na + .

| Cumulative meta-analysis
The cumulative meta-analysis showed no significant change over time for the effects of HSS plus furosemide versus furosemide alone on most outcomes (serum creatinine, serum Na + , and urine Na + ) (Supporting Information: Figures S4 and S5).  we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect. Abbreviations: BNP, type-B natriuretic peptide; CI, confidence interval; MD, mean difference; RCTs, randomized controlled trials; RR, risk ratio. a The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI

| Subgroup analyses
Subgroup analysis according to the type of population (refractory vs. unselected patients) showed that the interaction test was significant only for serum creatinine (p < .01) and serum Na + (p = .04) (Supporting Information: Table S3). Serum creatinine was significantly reduced only in refractory ADHF patients and serum Na + was significantly increased only in the same subgroup.
Subgroup analysis according to the duration of HSS plus furosemide (<6 vs. ≥6 days) showed that the interaction test was significant only for serum creatinine (p < .01) serum Na + (p < .01), although the effect remained significant in all strata (Supporting Information: Table S4).
Subgroup analysis according to the daily dose of furosemide (>200 mg/day vs. ≤200 mg/day) or country (Italy vs. non-Italian countries) showed that the interaction test was significant only for serum creatinine (p < .01) serum Na + (p < .01). Serum creatinine was significantly reduced only in patients treated with a daily dose of furosemide >200 mg/day and in studies performed in Italy. Serum Na + was significantly increased in all strata (Supporting Information: Tables S5 and S6).

| Sensitivity analysis
Sensitivity analysis including only trials with a low risk of bias showed that the results were consistent with the main analysis for all outcomes (Supporting Information: Table S7).

| DISCUSSION
This meta-analysis, including 10 RCTs and~3000 patients with HF, shows that treatment with intravenous HSS plus furosemide was associated with favorable responses across several surrogate efficacy endpoints, with no signals of safety based on low to moderate quality evidence (see Graphical abstract).
The administration of HSS is not used routinely in the management of ADHF patients. Current heart failure clinical practice guidelines dedicate too little attention to this therapy. 25 However, there is evidence that supports its efficacy and safety in ADHF patients. [15][16][17][18][19][20][21][22][23][24] The mechanisms underlying the effectiveness of HSS in decongesting patients with acute HF are diverse, including a rapid increase in plasma sodium and osmolality, with a rise in intravascular volume and renal perfusion, but are mainly focused on renal physiology. 26 A neurohormonal effect inhibiting the deleterious action of the renin-angiotensin system has been suggested. 27 This hypothesis has been supported by plasma determination of values of BNP and also other inflammatory and fibrotic parameters (suppression of tumorigenicity 2, inflammatory cytokines [IL-6]), which were lower in patients receiving HSS. 20,22 Diuretic resistance is a condition defined by an inability to increase fluid and sodium excretion despite an increase in loop diuretic dose, which is insufficient to relieve volume overload, peripheral edema, or pulmonary congestion. 28,29 The physiology of diuretic resistance is also complex and not fully understood. [28][29][30] It has been suggested that renal function (mainly glomerular filtration) would play a limited role in diuretic resistance, with sodium handling at the renal tubules being the most relevant mechanism. 30 Therefore, diuretic combination strategies targeting sodium reabsorption at different tubular levels are highly effective in patients with diuretic resistance, but at the cost of worsening kidney function and notable electrolyte abnormalities that harm patients' outcomes. 31 HSS administration improved the performance of loop diuretic therapy, as reflected in increased urinary volume and urinary sodium, and achieved a greater patient weight loss. 32 To that effect, HSS is an adjunctive useful measure in patients with diuretic resistance.
Furthermore, a trend towards a lower increase in creatinine levels was also observed in patients who received HSS. 21 Hyponatremia is a factor that markedly worsens the prognosis in ADHF patients, and its treatment can be complex. [33][34][35][36] Given the effectiveness of HSS in dealing with this problem, 37 In view of the available evidence, the following question should be raised: what could HSS offer compared to other conventional therapies for congestion in ADHF? First, fewer adverse effects were seen with HSS added to standard high-dose loop diuretic therapy, including fewer electrolyte abnormalities and lesser renal impairment.
Second, there are data that point to higher efficacy of HSS treatment, including a reduced length of hospital stay and faster resolution of congestion. Furthermore, the cost of HSS therapy is possibly a more cost-effective option than other ADHF therapies (including new diuretics or renal replacement therapy), and the shorter length of hospital stay would also contribute to a reduction in healthcare costs.
However, the administration of HSS in ADHF patients should not be undertaken indiscriminately, and it is necessary to properly select appropriate candidate patients that may potentially benefit from this therapy, for instance, those with serum sodium levels in the low range, or high risk of developing diuretic resistance. Based on the results of the intermediate outcomes, the use of HSS plus furosemide could be useful in ADHF patients with functional class NYHA III-IV and without response to initial diuretic therapy. The HSS concentration should be in accordance with the serum Na+ levels. Regarding the dose of furosemide, although several trials used high doses, the subgroup analysis did not show significant differences between doses higher and lower than 200 mg per day. Therefore, a lower dose could be used.
There are two previously published systematic reviews that evaluated the effect of HSS plus furosemide in ADHF patients (Supporting Information: Table S8). 43,44 Gandhi et al. 44 performed a meta-analysis of 10 RCTs published up to 2013. However, they erroneously pooled two related trials in which one was a follow-up to the other. 20,38 In addition, the risk of bias was assessed using the Newcastle-Ottawa Scale even though the tool was designed to be applied only in observational studies. In contrast, we conducted the risk of bias assessment using the most up-to-date version of the Cochrane tool for assessing RCTs. Covic et al. 43 performed a metaanalysis of 12 studies published up to 2020, combining data from observational studies and RCTs. The combination of these two study designs is not recommended because they could potentially lead to misleading results. Instead, our review focused only on fully published RCTs. Furthermore, we performed a trial sequential analysis to assess whether our results were sufficiently powered to produce firm conclusions about the efficacy of HSS plus furosemide.
In fact, our review is the only one that evaluated the quality of evidence for all outcomes using the GRADE approach. Overall, the relevance of our meta-analysis lies in that we used state-of-the-art methods to yield reliable conclusions overcoming the methodological problems of previous reviews.
This study has some limitations that should be acknowledged.
First, meta-analysis of all-cause mortality, cardiovascular mortality, readmission for heart failure was not conducted, and only a narrative synthesis was performed. Although none of included RCTs were adequately powered to evaluate these clinical outcomes, they showed a signal reduction in mortality and readmissions for heart failure. However, proper evaluation of these "hard" outcomes in future studies is necessary before providing a recommendation on their use in clinical practice since this is a critical aspect for other pharmacological agents (e.g., seleraxin, ularitide, nesiritide, or milrinone) previously studied in patients with ADHF. 45 Second, many studies included in our review come from a single healthcare system from on country (Italy). However, the meta-analysis of non-Italian trials was consistent with the main analysis. Third, although some studies included a small number of patients, the TSA-adjusted results showed firm evidence of benefit. Fourth, the observed benefit on hospitalization time may have been overestimated considering that most of the studies conducted in Italy presented longer hospitalization time at baseline. In addition, detailed information on the site (emergency department or hospital ward) where the patients were treated is not available in most studies in Italy. However, the direction of the effect is consistent among the included trials and the subgroup analysis showed no significant difference on hospitalization time when the Italian studies were excluded. Fifth, between-study heterogeneity was important for most outcomes. This may be explained by the type of population evaluated (different severity of clinical congestion), different timing of outcome assessment, and heterogeneous HSS protocol. This component was considered into the GRADE evaluation. Finally, most studies included patients with heart failure and reduced LVEF (<40%). It would be of interest to analyze in the future whether our findings are also applied to patients with heart failure and preserved ejection fraction.

| CONCLUSIONS
Our meta-analysis suggests that HSS plus furosemide compared to furosemide alone improves surrogate outcomes in ADHF patients based on low to moderate quality of evidence. However, further adequately powered RCTs are still needed to confirm our findings and to assess the effect on heart failure readmissions and mortality.