Systematic Review on the Management of Diuretic Resistance in Acute Heart Failure across the Spectrum of Kidney Disease Open Access

The incidence and prevalence of heart failure (HF) is increasing worldwide [1]. In addition to a high risk of mortality, this condition is associated with important morbidity, especially due to a high hospitalization rate [2, 3]. In contrast to the advances in the treatment of patients with stable chronic HF, the multiple interventions tested in patients with acute HF (AHF), especially those limited to the acute setting, have yielded disappointing results [4, 5]. Nevertheless, our understanding of the pathophysiology of AHF has increased, and congestion, defined as fluid accumulation due to elevated cardiac filling pressures, is not only the most common clinical finding that leads to an unplanned admission but also an important driver of poor prognosis [6, 7]. Accordingly, achieving decongestion has become an important goal in the management of patients with AHF [2, 5, 8]. In this regard, strategies that promote diuresis (and natriuresis), with loop diuretics as the backbone of treatment, have become appealing interventions and have been tested in large clinical trials [2].

Although the importance of adequate decongestion has been widely emphasized and different non-invasive tools have been developed to assess volume status, up to 50% of patients are discharged with clinical signs and symptoms of residual congestion, even in the context of randomized clinical trials (RCT) [6]. Many factors can explain these findings, such as difficulties in volume status assessment, inherent costs associated with prolonged hospitalizations, shortage of hospital beds, and the fact that a substantial number of patients exhibit worsening renal function (WRF) or diuretic resistance during the course of hospitalization [9‒11].

Diuretic resistance, defined as the inability to achieve euvolemia due to impaired response to diuretics, is subject to variable definitions and is not accurately defined in HF [2, 9]. Patients with AHF, compared with healthy individuals, are known not only to need higher doses of loop diuretics to reach the natriuretic threshold but also show a lower ceiling effect, meaning that higher diuretic doses are required for natriuresis with a lower net sodium excretion compared with healthy individuals [8]. The presence of chronic kidney disease (CKD), defined as abnormalities of kidney structure or function present for >3 months, a frequent comorbidity among patients with HF, further impacts diuretic efficacy [12, 13]. In fact, CKD is the most important risk factor for diuretic resistance [14, 15]. CKD is associated with impaired diuretic secretion into the tubular lumen and structural changes along the nephron that make patients more salt avid [8, 16, 17]. Altogether, these adaptative changes shift the natriuretic dose-response curve downward and to the right, making it more challenging to achieve decongestion [8].

As mentioned above, several clinical trials have studied different interventions to promote decongestion in patients with AHF and signs or symptoms of volume overload. The efficacy and safety of these treatment strategies in patients with overt or at risk of diuretic resistance, especially in the setting of CKD, is still incompletely described. The objectives of this systematic review are to summarize and critically analyze the evidence on decongestive treatment strategies in patients with HF presenting with either established or at risk of diuretic resistance across the spectrum of CKD.

This systematic review was conducted and reported according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) checklist (online suppl. Table 1; for all online suppl. material, see https://doi.org/10.1159/000546520) [18]. The study protocol was registered prior to data extraction and analysis in PROSPERO (registration number: CRD 42022375154).

We performed a systematic review of the literature using the PubMed/MEDLINE and Cochrane databases from inception to January 5, 2024. The search strategy is detailed in online supplementary Table 2. Only articles in English or French were considered, otherwise no restrictions were applied regarding publication date or status. In addition, the reference list of eligible studies, previous review articles, and systematic reviews/meta-analyses were reviewed, and a weekly email alert was used until March 30, 2024, to include RCTs published after January 5, 2024 (Fig. 1).

Articles that fulfilled the following inclusion criteria were included. (a) Study population: inclusion of patients with acute HF and either established or at risk of diuretic resistance, the last being defined as the presence of a baseline estimated glomerular filtration rate (eGFR) <60 mL/min/1.73 m². For trials not presenting a subgroup analysis of patients with baseline eGFR <60 mL/min/1.73 m², only those with a baseline mean + 1 standard deviation (SD) or 75th percentile eGFR <60 mL/min/1.73 m² were included. (b) Intervention: trials testing different pharmacological or invasive modalities compared to standard of care, placebo or an active comparator were considered, including intravenous loop diuretics, diuretic associations, vasoactive agents or ultrafiltration. Trials including therapies not currently commercially available were excluded. (c) Study design: only RCTs published in the form of a manuscript were considered. (d) Outcomes: at least one of the following outcomes had to be reported: (I) decongestion-related outcomes: metrics of diuretic efficiency, based on urine output, fluid balance or weight changes during the intervention; evaluation of signs or symptoms of decongestion, either patient-reported or based on scores that integrate clinical and imaging-derived markers or length of hospital stay; (II) safety outcomes: WRF, including the need for kidney replacement therapy, and electrolyte abnormalities, notably hypo- and hyperkalemia, hypo- and hypernatremia, and hypomagnesemia; and (III) clinical outcomes: rates of all-cause, cardiovascular death or hospitalization at 90 days.

We considered decongestion-related outcomes as the primary outcome of this systematic review given the known association of decongestion with hard clinical endpoints and the fact that most trials were not powered to ascertain an effect on mortality or heart failure hospitalizations. The safety and clinical outcomes stated above were considered as secondary outcomes of our analysis.

Two authors (P.M. and M.B.H.) independently reviewed the literature and selected the studies for full-text review. If disagreements on study selection were identified, the senior author (T.A.M.) was consulted.

Trial data were extracted using a standardized digital form. The following information was systematically retrieved from each trial: article identification, including the trial name, first author’s name, and year of publication; information about study design, including inclusion and exclusion criteria, time between admission and randomization, sample size, intervention and comparator, key baseline characteristics; and data on decongestion- and safety-related outcomes. The risk of bias was assessed with the RoB2 tool [19] (online suppl. Fig. 1).

Given the inclusion of two different groups of patients, those with established diuretic resistance and at risk of diuretic resistance, the result section will describe the included studies separately, for clarity. We report decongestion-related outcomes and safety outcomes, as per our study protocol. In addition, we are reporting short-term clinical endpoints, including death or hospitalization at 90 days.

As prespecified in the registered protocol, a meta-analysis of the most commonly reported outcomes was considered. However, given the heterogeneity of trial design, intervention strategies, patient population, and selected outcomes, a decision not to pool the results was made.

A total of 4,020 articles were identified and screened. Eighty-six trials were selected for full-text review and 22 RCTs were finally included. Of those, 6 trials enrolled 529 patients with established diuretic resistance (Table 1) and 16 trials a total of 1,913 patients at risk of diuretic resistance (Table 2). The study flowchart is shown in Figure 1.

Six RCTs that recruited patients based on the presence of diuretic resistance were included in our systematic review (Table 1). All trials enrolled patients presenting with AHF. The K-STAR trial also included patients with chronic HF and signs or symptoms of congestion [23]. The diuretic resistance criterion was different across trials but was invariably based on the presence of signs/symptoms of congestion unresponsive to a defined dose of loop diuretics or the inability to induce a total urine volume or weight loss above a predefined cutoff. All included trials tested pharmacological agents that sequentially blocked different sites of the nephron on top of loop diuretics to promote decongestion (Fig. 2). They were either compared with placebo or an active treatment arm. All trials had a relatively small sample size, ranging from 27 to 250 patients in total. The number of patients screened for inclusion to the study was only reported in the DAPA-RESIST trial (1,651 screened for a total of 61 enrolled patients). Across all included studies, baseline mean or median eGFR was consistently <60 mL/min/1.73 m². The time from hospital admission to randomization was not provided, with the exception of the DAPA-RESIST trial (6 days, interquartile range 4–11 days).

Channer et al. [20] compared two different thiazide diuretics, bendrofluazide and metolazone (10 mg each on top of 160 mg/day of intravenous furosemide), and two different treatment periods (either for 3 days or at the clinician's discretion), using a 2 × 2 factorial design in 33 patients. Diuretic resistance was defined as the absence of weight loss for ≥2 days while on 160 mg/day of intravenous furosemide. Both bendrofluazide and metolazone were equally effective in inducing diuresis and weight loss without severe electrolyte imbalances. The authors also found that a 3-day regimen was equally effective compared with a more prolonged treatment course (median 5.6 days).

Hanatani et al. [21] enrolled 44 patients admitted with AHF and diuretic resistance, defined as the persistence of signs/symptoms of congestion despite receiving conventional diuretic therapy (a loop diuretic equivalent to ≥40 mg of oral furosemide alone or in combination with a thiazide or an aldosterone antagonist at any dose). Patients were randomized in a 1:1 ratio to receive a 7–14 days course of tolvaptan 15 mg daily with either the same loop diuretic dose as in the 3-day run-in period or with half the daily loop diuretic dose. Both treatment strategies were equally effective in achieving decongestion (measured as daily urinary output and weight loss). The reduced loop diuretic dose arm had a lower incidence of WRF with a mean change in creatinine of −0.07 ± 0.19 mg/dL, while the fixed-dose loop diuretic group showed a mean creatinine change of +0.14 ± 0.08 mg/dL (p = 0.006). Although statistically significant, the authors stated that the clinical relevance of this difference is debatable.

In the SECRET of CHF trial, 250 patients with AHF were randomized to receive either tolvaptan 30 mg/day or placebo for up to 7 days [22]. For enrollment, one of the following criteria was also required: hyponatremia, renal dysfunction, or diuretic resistance. Diuretic resistance was defined as a urinary output of ≤125 mL/h in any ≥2-h period in the 8 h following the administration of an intravenous loop diuretic equivalent to ≥40 mg of intravenous furosemide. The authors did not specify the overall proportion of patients included with the diuretic resistance criterion, but 30 patients (12%) were included based on this criterion alone. Tolvaptan led to a significantly greater weight loss at 24 h compared with placebo, without affecting the primary endpoint of dyspnea improvement. No results were provided in the subgroup of patients with diuretic resistance.

In the K-STAR trial, 81 patients with acute or chronic HF and fluid retention despite taking ≥40 mg/day of furosemide equivalents were randomized to receive either ≤15 mg of tolvaptan or an increased loop diuretic dose for 7 days. Addition of tolvaptan resulted in a significant increase in mean daily urine output (459 ± 514 mL vs. 79 ± 341 mL, p < 0.01) and in total urine output during the treatment period. However, there was no significant effect on weight loss (−2.1 ± 1.8 kg vs. −2.1 ± 2.6 kg, p = 0.61). The incidence of WRF, defined as an increase in SCr ≥0.3 mg/dL during the treatment period, was 18% in the tolvaptan group vs. 44% in the increased loop diuretic dose group (p = 0.01) [23].

In the 3T trial, diuretic resistance was defined as a 12-h urine output of <2 L despite receiving ≥240 mg/day of intravenous furosemide equivalents in 12 h [24]. A total of 60 patients with AHF meeting the diuretic resistance criterion were randomly assigned to receive 5 mg of oral metolazone twice daily, 500 mg of intravenous chlorothiazide twice daily or 30 mg of tolvaptan orally once daily, on top of high-dose intravenous loop diuretics. All interventions equally improved weight-based diuretic efficacy (defined as weight loss/40 mg of intravenous furosemide) but tolvaptan also significantly improved urine output-based diuretic efficacy (defined as urinary output/40 mg of intravenous furosemide), compared with oral metolazone. The study primary outcome, weight loss at 48 h, was not significantly greater with either intravenous chlorothiazide or tolvaptan compared with oral metolazone. Moreover, the total urine sodium excretion at 48 h as well as the spot urine sodium at 48 h were significantly lower with tolvaptan compared with both oral metolazone and intravenous chlorothiazide (48-h total urinary sodium excretion: 638 ± 316 mmol vs. 976 ± 341 mmol and 1,195 ± 330 mmol, respectively [p = 0.02]; 48-h spot urinary sodium: 58 ± 25 mmol/L vs. 104 ± 16 mmol/L and 117 ± 14 mmol/L, respectively [p < 0.01]).

In the most recent DAPA-RESIST trial, 61 patients with AHF and diuretic resistance were randomized to receive a 3-day course of metolazone or dapagliflozin [25]. Diuretic resistance was defined as <1 kg weight loss or <1 L negative fluid balance in 24 h, despite ≥160 mg/day of intravenous furosemide. Both treatments led to similar weight loss at 96 h, the primary outcome, but dapagliflozin was associated with a numerically lower diuretic efficiency (weight loss/40 mg of intravenous furosemide). Metolazone led to a statistically significantly higher urinary sodium excretion at all time points compared with dapagliflozin, but this did not result in any significant differences in daily urine output and cumulative net fluid balance. There were also no significant differences in other metrics or decongestion scores (B-lines on lung ultrasound, pleural effusion score or modified ADVOR score) or overall length of stay between the two groups.

A total of 16 RCTs met the inclusion criteria. The main characteristics of the trials are shown in Table 2. As expected, most strategies involved blocking different sites of action along the renal tubules (Fig. 2). All decongestion promoting strategies will be presented according to the pharmacological category they belong to. From the RCTs described in this section, in 12 trials, data are reported for the overall population enrolled, while for the DRAIN, UNLOAD, ATHENA-AHF, and DICTATE-AHF trials only the subgroup of patients with kidney dysfunction at baseline was considered [27, 30, 33, 36].

Only two RCTs testing different loop diuretic strategies in AHF included a population at risk of diuretic resistance [26, 27]. Both trials compared a strategy of continuous infusion vs intermittent intravenous furosemide.

In the diurHF trial, in 57 patients with AHF, an ejection fraction <45%, and a mean baseline eGFR of 45 mL/min/1.73 m², the continuous infusion arm was associated with greater daily urinary output and better decongestion (as measured by the reduction in B-type natriuretic peptide [BNP] levels) [26]. The continuous infusion arm resulted in a greater drop in eGFR and a higher incidence of acute kidney injury (AKI), defined as an increase in serum creatinine >0.3 mg/dL. Moreover, the continuous furosemide infusion arm led to significantly longer length of stay (14.3 ± 5 days vs. 11.5 ± 4.3 days, p < 0.03) and a higher incidence of the composite outcome of hospitalization for heart failure or all-cause mortality at 6-months (15 [43%] vs. 9 [34%], p < 0.03). In addition, the continuous infusion arm received more frequently intravenous dobutamine during the hospitalization (50% vs. 26%, p < 0.01). The authors state that the potentially worse hemodynamic state in this treatment arm could have resulted in the discrepancies seen in decongestion-related and hard clinical endpoints. Nevertheless, given the small number of events, the results of this trial are hard to interpret.

In the DRAIN trial, in the subgroup of patients with an eGFR <30 mL/min/1.73 m² at baseline, continuous furosemide infusion was associated with numerically higher odds of freedom from congestion at 72 h, defined as jugular venous pressure of <8 cm, no orthopnea and no or only trace peripheral edema. However, the interpretation of these results is limited by the small sample size (n = 21) [27].

The assumption that the proximal renal tubule is responsible for a high proportion of sodium reabsorption, especially in HF, led to trials testing the efficacy of the carbonic anhydrase inhibitor acetazolamide in AHF (Fig. 2). In the DIURESIS-CHF trial (n = 34), intravenous acetazolamide (at a dose of 250–500 mg) on top of intravenous loop diuretics was compared with intravenous loop diuretic monotherapy at twice the oral maintenance dose in patients at risk for diuretic resistance [28]. Natriuresis at 24 h was similar in both treatment arms. Therefore, a higher loop diuretic efficacy (i.e., natriuresis corrected for loop diuretic dose) was observed in the acetazolamide arm.

These encouraging results led to the design of the multicentric ADVOR trial. Patients with AHF were randomized to receive either 500 mg of intravenous acetazolamide or placebo [29]. In this study, >80% of patients enrolled had an eGFR <60 mL/min/1.73 m². The primary outcome, successful decongestion at 72 h, defined as absence of signs of volume overload (identified by means of clinical examination and ultrasound assessment of pleural effusion/ascites) and no indication for escalation of decongestive therapy, was more frequently observed in the acetazolamide arm (risk ratio 1.46 [95% CI: 1.17–1.82] favoring acetazolamide). Although the results were more marked in patients with baseline eGFR below the median (<39 mL/min/1.73 m²), patients receiving a higher home loop diuretic maintenance dose appeared to have less benefit than those treated with lower maintenance dose (>60 mg vs. ≤ 60 mg of furosemide equivalents: 1.78 [95% CI: 1.33–2.36] vs. 1.08 [95% CI: 0.76–1.55], respectively). With the exception of a shorter length of hospital stay, acetazolamide was not associated with any benefit in the other prespecified or exploratory outcomes.

The growing evidence of benefit of sodium-glucose co-transporter 2 inhibitor (SGLT2i) in HF together with the potential for enhanced decongestion with these agents led to multiple trials testing their role in AHF [43‒45]. However, only the DICTATE-AHF trial provided results on decongestion-related outcomes in a group of patients at risk of diuretic resistance based on the presence of low baseline eGFR. The primary endpoint was mean weight change per 40 mg furosemide equivalents and it was also reported in the subgroup of patients with an eGFR below the median value (eGFR <51–54 mL/min/1.73 m², n = 120) [30]. In this subgroup, as in the overall study population, dapagliflozin did not result in a significantly higher diuretic efficiency, compared with usual care (−0.30 vs. −0.28 kg/40 mg furosemide equivalents). Diuretic efficiency in patients treated with dapagliflozin (vs. usual care) was lower in those with a baseline eGFR below the median, compared with those with an eGFR above the median (p value for interaction 0.08).

The CLOROTIC trial assessed the added value of hydrochlorothiazide in patients with AHF and kidney dysfunction at baseline (defined as an eGFR <60 mL/min/1.73 m²) [31]. Hydrochlorothiazide enhanced diuretic response at 72 h but did not result in improvements in patient-reported dyspnea, compared with placebo. Hydrochlorothiazide was associated with a higher incidence of WRF, defined as an increase in SCr >0.3 mg/dL or a >50% decrease in eGFR. Most of the WRF events were due to a mild increase in SCr >0.3 mg/dL from baseline, with only 1 patient in each group experiencing a 50% decrease in eGFR. There was no difference in all-cause mortality and hospital admissions at 30 days.

The use of MRAs in patients presenting with AHF at risk for diuretic resistance has been limited to spironolactone [32, 33, 46]. In the DIURESIS-CHF trial, patients with AHF and renal dysfunction were randomly assigned to receive 25 mg of spironolactone upfront or upon discharge [32]. Although the study only managed to include 34 out of 80 planned patients, upfront use of spironolactone was associated with significantly higher 24-h natriuresis (314 ± 142 mmol/L vs. 200 ± 91 mmol/L, p = 0.01). A numerically lower incidence of potassium abnormalities, including hyper and hypokalemia, was seen with upfront spironolactone (19% vs. 39%, p = 0.27).

In the multicentric ATHENA-AHF trial, a natriuretic dose of spironolactone (100 mg) was compared with placebo and standard of care (spironolactone 25 mg) for a duration of 4 days in patients admitted with AHF [47]. In a post hoc analysis, Greene et al. [33] analyzed the results of this trial within tertiles of baseline eGFR. In the lowest eGFR tertile (<50 mL/min/1.73 m²), 60 patients received high-dose spironolactone and 58 were randomized into the standard of care/placebo arm. Patients in the lower eGFR tertile had the higher N-terminal pro B-type natriuretic peptide (NTproBNP) levels at baseline and the lowest likelihood of experiencing decongestion during the acute episode, as measured by the 96-h change in log NTproBNP levels or by congestion/dyspnea scores. Nevertheless, those who were randomized to receive high-dose spironolactone, showed a more pronounced decrease in NTproBNP levels and numerically higher urine output and weight loss at 96 h, compared with placebo or standard of care. This suggests a potential beneficial effect of high-dose spironolactone in this subgroup. Importantly, high-dose spironolactone was found to be safe and well tolerated, without a significant drop in eGFR or higher incidence of hyperkalemia compared with placebo or standard of care.

Vasopressin is implicated in water retention in HF and two clinical trials examined the potential role of V2 receptor antagonists in patients at risk of diuretic resistance, based on the presence of kidney dysfunction at baseline [34, 35]. In the AQUAMARINE trial, 217 patients with an eGFR <60 mL/min/1.73 m² were randomly assigned to receive either tolvaptan or standard of care [34]. Tolvaptan significantly increased the 48-h urine output, the study primary outcome, with a 1.5 L mean difference between the two groups. Results were similar in a subgroup analysis of patients with baseline eGFR <30 mL/min/1.73 m². Tolvaptan was also associated with increased weight loss, negative net fluid balance, higher rates of dyspnea improvement, as well as a lower cumulative diuretic dose.

Komiya et al. [35] enrolled 33 patients with HF and advanced CKD (eGFR <30 mL/min/1.73 m²) after a 3-day run-in period during which they were treated with 20–100 mg/day of furosemide. Patients were then randomized to receive tolvaptan 15 mg/day on top of the same loop diuretic regimen or an increase in daily furosemide dose of 100 mg for up to 7 days. The mean pre- to post-randomization change in daily urine output significantly increased in the tolvaptan arm (637 vs. 119 mL/day). This was accompanied by a significantly lower rate of WRF (defined as a ≥0.3 mg/dL increase in SCr), compared with the high-dose loop diuretic arm. However, despite these encouraging results, vasopressin receptor antagonists are not used routinely in acute heart failure after the EVEREST trial failed to show any benefit with respect to cardiovascular mortality or hospitalization for heart failure in this population [48].

Use of ultrafiltration to induce a negative fluid balance in patients with AHF was compared with standard of care or a standardized diuretic protocol in four clinical trials [36‒39]. Only in the CUORE trial (n = 56), use of intravenous loop diuretics was allowed simultaneously with ultrafiltration and the daily loop diuretic dose during hospitalization was similar in both study arms. In this trial, patients allocated to ultrafiltration had similar weight loss and changes in BNP levels upon discharged comparing with standard of care [39]. In addition, there were no significant in-hospital differences in kidney function between treatment arms. Lastly, ultrafiltration did not impact in-hospital length of stay.

In the ULTRADISCO trial (n = 30), ultrafiltration led to more pronounced decongestion at 36 h, both from a clinical (weight loss and NYHA class) and biomarker (NTproBNP reduction) point-of-view, compared with loop diuretics [37]. In the UNLOAD trial, ultrafiltration was associated with greater weight loss at 48 h compared with standard of care [36]. Although the results for the primary outcome were not reported in the subgroup of patients with a baseline SCr >1.5 mg/dL (n = 67), the authors state that no heterogeneity was identified with respect to the effect of ultrafiltration on weight loss at 48 h.

Finally, in the landmark CARESS trial, ultrafiltration led to similar weight loss compared with a standardized diuretic up-titration protocol [38]. This was accompanied by a significantly higher increase in mean serum creatinine at 96 h and a higher incidence of serious adverse events (including WRF, bleeding, and catheter-related complications).

Two clinical trials from the 1990s including ≤10 patients per treatment arm studied the effect of dopamine, a catecholamine that in low doses is considered to act primarily as a renal vasodilator, on decongestion and renal function in AHF [40, 41]. The studies had conflicting results (Table 2).

The multicentric ROSE trial recruited patients with AHF and renal dysfunction (eGFR 15–60 mL/min/1.73 m²). A total of 122 patients were randomized to the low-dose dopamine arm and 119 to placebo [42]. In this trial, dopamine failed to impact the decongestion endpoint, the cumulative urinary output at 72 h. Similarly, dopamine did not show any benefit in any of the prespecified secondary endpoints, including patient-reported dyspnea or diuretic efficiency. Regarding safety endpoints, dopamine did not impact the renal function co-primary endpoint, defined as the change in serum cystatin C levels at 72-h post-randomization. Similar results were seen in the subgroup of patients with an eGFR <44 mL/min/1.73 m².

Our systematic review in acute heart failure focusing on patients with established or at risk of diuretic resistance has identified the following key findings: (1) although the definition of diuretic resistance lacks standardization, its incidence appears to be low in those who receive appropriately high doses of loop diuretics; (2) while most studies were designed to achieve decongestion while avoiding WRF, most trials lack data on a pre-existing baseline value at steady state and still report minimal changes in creatinine or cystatin C levels as safety outcomes that may not accurately reflect clinically significant kidney injury; (3) despite the efficacy of the various strategies reported in our systematic review in promoting decongestion, no major clinical impact was seen post-discharge. In this regard, initiation of therapies known to positively impact hard clinical outcomes in patients with chronic HF should form, along with intravenous loop diuretics, the mainstay of treatment during the in-hospital period.

Although diuretic resistance is reported to be common in HF, its definition lacks standardization and a clear correlation with hard clinical endpoints has not been demonstrated [11, 49, 50]. RCTs that included patients with diuretic resistance used markedly variable criteria to define this entity. In addition, the number of patients included in these trials was small, suggesting that diuretic resistance may be less common than initially thought. Consistently, in the DAPA-RESIST trial, out of 1,651 patients screened, 1,587 patients (96%) were excluded, at least 621 (40%) due to absence of diuretic resistance [25]. Similarly, in the SECRET of CHF trial, although the total number of patients screened was not provided, 37 different sites and 4 years were required to enroll a total of 250 patients, suggestive of a slow recruitment rate [22]. Finally, trials such as the PURE-HF trial (NCT 03161158) or the Metolazone Versus Chlorothiazide for Acute Decompensated Heart Failure With Diuretic Resistance (NCT 03574857), aiming to recruit patients with AHF and diuretic resistance, were either terminated early or discontinued due to slow recruitment.

A potential explanation for the low incidence of diuretic resistance is the use of appropriate loop diuretic regimens in most recent studies. Indeed, after the DOSE trial, a standardized and stepped diuretic algorithm was used for patients presenting with AHF, using at least the double dose of intravenous diuretics upfront [17, 29, 38, 42, 51]. The strong diuretic response seen with this approach in trials like CARESS, ROSE, or ADVOR, when compared with ultrafiltration, dopamine/nesiritide, or acetazolamide, respectively, suggests that most patients classified as diuretic resistant in clinical practice may in fact be treated with inadequate doses of intravenous loop diuretics. The significant interaction seen in patients treated with lower vs higher home maintenance doses of loop diuretics in the ADVOR trial further supports this statement [29].

Another important caveat with current definitions of diuretic resistance relates to our inability to identify these patients early in the course of the hospitalization. In the DAPA-RESIST trial, patients were randomized 6 (4–11) days post hospital admission, demonstrating our inability to recognize treatment resistance early. Moreover, current definitions rely on urinary output, net fluid balance or weight loss metrics that may be difficult to obtain consistently in clinical practice, often due to variability in documentation and clinical workflows. Furthermore, these metrics were only correlated with hard clinical outcomes in observational studies or post hoc analyses [9, 52].

A machine learning analysis of pooled patient-level data from the DOSE, ROSE-AHF, ATHENA-HF, CARESS-HF, and ESCAPE trials identified 3 phenotypic groups based on diuretic efficiency, defined as urinary output divided by loop diuretic dose within 72 h of enrollment [53]. Using random forest-derived variables associated with the lowest efficiency group, the BAN-ADHF score (history of hypertension, atrial fibrillation, recent HF hospitalization, home diuretic dose, diastolic blood pressure, serum creatinine, blood urea nitrogen, and NTproBNP levels) was developed. In external validation, a higher BAN-ADHF score correlated with lower diuretic efficiency, worse clinical outcomes, longer hospital stays, and increased in-hospital mortality, suggesting its potential as a tool for early identification of diuretic resistance and guiding more intensive management and closer follow-up during the decongestive phase. Another potentially interesting approach to identify patients with diuretic resistance in AHF is the furosemide stress test (FST). A low urine output (<200–300 mL) in the 2–3 h after a 1–1.5 mg/kg bolus of intravenous furosemide is an established marker in critical care for patients presenting with early AKI and has been shown to predict progression to stage 3 AKI or short-term all-cause mortality, outperforming known markers of renal tubular damage [54, 55]. In a prospective study including 65 patients with AHF, low 2-h natriuresis following the FST, defined as a spot urinary sodium below the median (<113 mmol/L), was associated with lower diuretic response, based on ≤30% decrease in NTproBNP or composite congestion score<3 at day 5 [56]. Furthermore, patients with a low natriuresis following the FST also showed longer hospitalization periods, increased rated of hospital readmission and higher risk of mortality. The FST might, therefore, allow for the timely identification of diuretic resistance especially in patients with cardiorenal syndrome-related AKI or advanced CKD at baseline. These points highlight how the absence of a standard and clinically applicable definition of diuretic resistance precludes these patients from being properly identified thereby preventing the development of treatment strategies to improve their prognosis. We propose that future trials integrate the BAN-ADHF score or the FST to define diuretic resistance and guide clinical practice in this population.

Another important point that merits discussion is the absence of data, in most studies, on past medical history of CKD or a stable baseline eGFR outside the acute event. The rare exceptions were the ATHENA-AHF and EVEREST trials [33, 48]. While the prevalence of CKD at baseline was 27% in both trials, post hoc subgroup analyses showed that more than 50% of participants had an eGFR <60 mL/min/1.73 m² upon enrollment [33, 57]. This suggests that up to 50% of patients with AHF who present with an eGFR <60 mL/min/1.73 m² may in fact have acute cardiorenal syndrome. It is plausible that these patients may not have the underlying advanced structural changes or loss of functional nephrons that would make them unresponsive to decongestion therapies and, therefore, are not diuretic resistant [8, 58]. Illustrative of this, few patients within the lower tertile of baseline eGFR in the ATHENA-AHF trial (eGFR 30–44 mL/min/1.73 m²) or below the median eGFR in the ROSE study (eGFR <44 mL/min/1.73 m²) showed signs of diuretic resistance [33, 42]. In conclusion, the presence of kidney dysfunction alone, especially in the setting of acute cardiorenal syndrome, may not identify an AHF subpopulation that will be diuretic resistant. The lack of detailed information on baseline renal function precludes the understanding of the impact of CKD on decongestion and hard clinical endpoints in patients with AHF. It is important that future trials collect data on baseline eGFR as a clinical parameter.

It is interesting to note that most strategies presented in this systematic review were developed with the goal of promoting decongestion while preserving kidney function. However, it has been shown that deterioration in kidney function during the decongestion phase of AHF is not associated with an increase in tubular injury biomarkers or worse clinical outcomes [59‒61]. If coupled with proper decongestion, patients with WRF at discharge have in fact a better prognosis compared to those with stable renal function and residual congestion [62]. Moreover, changes in kidney related biomarkers during the course of the hospitalization should consider the complex clinical context of the individual patient, including his clinical status, diuretic response, initiation of guideline directed medical therapy or non-HF-related complications that may occur during the in-hospital stay [63]. Despite this, contemporary trials still report an increase of 0.3 mg/dL in SCr as a safety outcome [25, 29, 31]. We believe that the excessive emphasis on clinically irrelevant changes in kidney related biomarkers in AHF trials perpetuates clinical practices that are too much focused on preserving kidney function, which may ultimately lead to incomplete decongestion and worse hard clinical outcomes. This acute form of “renalism” resembles wrong preconceptions that preclude CHF patients with CKD from receiving prognosis modifying therapies, such as renin-angiotensin-aldosterone antagonists or SGLT2i [64‒67].

The use of urinary sodium concentration in a spot urine sample has recently gained a renewed interest for the management of decongestion strategies in AHF. Using urinary sodium to guide loop diuretic dose and readily identify treatment resistance is of potential interest, especially in patients at risk of diuretic resistance. Expert opinions and more recently pragmatic trials tested this strategy and suggested that a spot urinary sodium-based approach resulted in higher natriuresis and diuresis and, in the ENACT-HF trial, also led to a shorter length of stay [68, 69]. Although considered easily applicable in clinical practice, trials using urinary sodium measurements were limited to the first 36–48 h of hospitalization. Furthermore, for all metrics of diuretic efficiency, this approach was not superior to the standard of care when both groups were compared at 72 h in the PUSH-AHF trial [69]. In fact, decongestion, either based on clinical parameters or congestion scores, is a challenging outcome to achieve. In the CARESS trial, after 96 h of treatment with either ultrafiltration or a standardized stepped diuretic protocol that included sequential nephron blockade with a thiazide, less than 10% of patients experienced clinical decongestion in both arms [38]. In the DAPA-RESIST trial, the mean congestion score was still high in both the dapagliflozin and metolazone arms at 96-h post-randomization and the overall median length of stay was close to 20 days in both treatment arms [25]. Therefore, it is questionable whether a urine sodium-based approach limited to the first 48 h will impact decongestion throughout the hospitalization period, especially as diuretic efficiency is known to gradually decrease over time [8, 70].

In the trials included in this systematic review, although the various treatment strategies tested were equally efficient in promoting decongestion in patients with or at risk for diuretic resistance, there was no significant impact on hard clinical endpoints. This finding might be due to the short duration of most of the interventions tested and our inability to properly assess and achieve decongestion. In clinical practice, we propose the following strategy (Fig. 3, central illustration): upfront use of high-dose intravenous loop diuretics is an essential step for achieving decongestion in this population. No compelling data suggest any added benefit of a continuous loop diuretic strategy vs bolus, and it might in fact be associated with more hemodynamic instability and neurohormonal activation [26, 71]. In addition to decongestion, clinicians should focus on using treatments known to impact hard clinical outcomes in patients with chronic HF, such as MRA and SGLT2i [47, 72]. Along with these agents, initiation of angiotensin receptor neprilysin inhibitors, compared with enalapril, has also been shown to impact outcomes post-discharge in patients with HF and reduced ejection fraction [73]. The safety and tolerability of initiating these interventions during the acute inpatient setting has been shown even in patients with lower baseline eGFR [30, 47, 73]. Altogether, based on data from clinical trials, this strategy is expected to be effective in promoting decongestion in >90% patients. Tools, such as the use of urinary spot sodium measurements, could be used to monitor the diuretic response during the decongestion phase and, together with the BAN-ADHF score and FST, rapidly identify patients with inadequate diuretic response and formerly establish diuretic resistance. In those cases, uptitration of loop diuretics or diuretic enhancing strategies, such as the use of thiazide diuretics, spironolactone, or acetazolamide could be considered to further promote diuresis/natriuresis. The use of a combination of diuretics (thiazides and MRAs on the top of furosemide) is a strategy that seems to have a similar efficacy to high-dose furosemide [74]. The use of acetazolamide could be particularly useful in those on lower home maintenance doses of loop diuretics or higher baseline serum bicarbonate levels [29, 75]. However, it should be noted that their combination with SGLT2i, who also primarily target the proximal renal tubule, has never been tested in RCTs. If, on the other hand, an increase in loop diuretics dose is the selected approach, it should be noted that diuretic response follows a logarithmic relation rather than linear, underlying the need for doubling the dose if no response is obtained [8]. In this regard, peak daily furosemide equivalent doses of 1,500 mg have previously been used without complications [76]. Finally, the use of congestion scores along with assessment of decongestion using non-invasive lung ultrasound, venous excess ultrasound or incorporating other digital derived biomarkers, holds promise in helping clinicians identifying residual congestion and effectively attaining decongestion, potentially translating into better clinical outcomes post-discharge [77‒81].

In conclusion, the incidence of diuretic resistance in AHF appears to be lower than previously described. Different management strategies can be considered in the acute setting, but use of high-dose intravenous loop diuretics is the cornerstone of management for this patient population. Instead of focusing on clinically insignificant small changes in serum creatinine from baseline, effective decongestion strategies should be prioritized in this highly morbid population to improve hard clinical endpoints.

The figures included in this manuscript were created using the BioRender^® software.

A statement of ethics is not applicable because this study is based exclusively on published literature.

Pedro Marques received salary support from an educational grant by Janssen. João Pedro Ferreira has received research support from Boehringer Ingelheim, AstraZeneca and Novartis, through his institution, the University of Porto. Michael A. Tsoukas reports personal fees from Boehringer Ingelheim, AstraZeneca, Janssen, Novo Nordisk, and Eli Lilly. Abhinav Sharma has received support from the Fonds de Recherche Santé Quebec (FRSQ) Junior 1 clinician scholars’ program, Canadian Institute of Health Research (Grant #175095), Roche Diagnostics, Boehringer Ingelheim, Novartis, Janssen, Novo Nordisk, Servier, AstraZeneca, and Takeda. Thomas A. Mavrakanas received speaker honoraria and has served on advisory boards for Bayer, Böhringer-Ingelheim, and Novo Nordisk outside the submitted work. He has also received a research grant from Astra Zeneca, Janssen, Pfizer, and Bayer. He is supported by an FRQS Clinician Scholar Award. The remaining authors have nothing to disclose.

No funding was received for this work.

P.M. was responsible for the design, query definition, literature search, and writing of the first and final version of the manuscript.

M.B.H. was responsible for the literature search (serving as a second reviewer) and reviewed the final version of the manuscript.

T.T., J.P.F., M.A.T., and A.S. reviewed the first version of the manuscript and provided valuable insights to the discussion.

T.A.M. was responsible for the design, query definition, final list of trials to be included in the systematic review; provided supervision; and reviewed all versions of the manuscript.

The data that support the findings of this study are not publicly available due to privacy reasons but are available from the corresponding author upon reasonable request.