Treatment of chronic heart failure (HF) with reduced ejection fraction (HFREF) is based on four main drug classes that have a prognostic impact, namely angiotensin-converting enzyme inhibitors (ACEIs) or angiotensin receptor/neprilysin inhibitors (ARNIs), beta-blockers (BBs), mineralocorticoid receptor antagonists (MRAs), and sodium-glucose cotransporter 2 inhibitors (SGLT2is).1

The use of loop diuretics in HF management is central when congestion is detected.1 Despite the evidence regarding loop diuretics in HF,2,3 their use remains largely empirical. A previous meta-analysis including studies with a small number of patients has shown an association between loop diuretic use and reductions in HF disease progression, as well as improvements in patients’ functional capacity.4 There is still, however, a clear lack of evidence concerning the impact of loop diuretics on HF prognosis. In HF management, loop diuretic dose is frequently increased for a period in order to reduce congestion. In euvolemic patients, diuretic dose reduction is recommended in various guidelines and position papers. Nevertheless, there are currently no recommendations concerning their discontinuation. Although this is a frequent decision in clinical practice, the underlying evidence is scarce and conflicting.5–8 Additionally, the recent indication for use of SGLT2is in HFREF may also have an impact on the need for diuretic maintenance, as does the use of ARNIs, which have been associated with the ability to maintain lower doses of loop diuretics.3,8–10 This link is particularly important, as loop diuretic use can lead to electrolytic changes, neurohumoral activation, worsening of kidney function and symptomatic hypotension, which may limit drug uptitration as part of guideline-directed medical therapy (GDMT).1,11

The aim of this study was to retrospectively assess the impact of loop diuretic discontinuation on the prognosis of euvolemic outpatients with HFREF. The primary endpoint was analysis of congestive events within one year of diuretic discontinuation.

The present study aimed to assess the impact of the discontinuation of loop diuretics on the prognosis of patients with HFREF. The traditionally hospital-based concept of WHF has shifted in recent years toward the inclusion of HF outpatients who require an increase in diuretics,12–14 but a consensus on its definition has yet to be reached.15 In our work, we considered all events that required an unscheduled reassessment and an increase in loop diuretics, even in an outpatient setting, in line with the latest definition of WHF in accordance with a position paper from the Heart Failure Association of the European Society of Cardiology.16

In our cohort of patients with HFREF, 265 were on loop diuretics, which it was possible to discontinue at some point in almost half (129 patients, 48.7%). Only 18 patients (14.8%) had a congestive event among the 122 patients with one year of follow-up, and 15 of these events (83.3%) were simply loop diuretic resumption at a scheduled visit, with a low diuretic dose. Therefore, only three (2.5%) WHF events occurred during the one-year follow-up. This clearly shows that in selected patients with HFREF, diuretics can be safely discontinued.

Patients in our trial were well treated in terms of GDMT: at discontinuation, 98.45% were on ACEIs/ARBs/ARNIs, 93.8% on BBs and 79.1% on MRAs. The number of patients on SGLT2is was low, as most data were collected prior to the publication of DAPA-HF,17 the first trial showing the benefit of SGLT2is in HFREF. It is also worth noting that our patients had a low median age, with low B-type natriuretic peptide (BNP) values and NYHA class, normal blood pressure and controlled heart rate, with creatinine, eGFR and potassium values within normal ranges. These patient characteristics may have enabled diuretic discontinuation.

Furthermore, GDMT was optimized from baseline to discontinuation, with significant increases in ARNIs (and thus decreases in ACEIs/ARBs), MRAs and SGLT2is. This, as expected, led to clinical and laboratory improvement, with a significant improvement in LVEF, along with significant reductions in heart rate, which are associated with improvement in clinical outcomes.18,19 It also enabled a significant reduction in furosemide doses, supporting the physician's decision to discontinue the drug. There was also a significant rise in potassium levels, possibly attributable to the significant increases in MRA use.20,21

Discontinuation of diuretics was dictated solely by assessing congestion with physical examination for most patients (natriuretic peptide [NP] data were available in only 61 patients). Although there are numerous tools available to assess congestion, as well as many different alternatives under investigation,3,22,23 it was interesting to note how safe discontinuation was possible in these patients using only semiology.

Significant improvements in kidney function were observed from discontinuation to one-year follow-up. This is an unexpected finding, as patients with HF experience significant deterioration in kidney function along their journey.24 A possible explanation is diuretic discontinuation, as SGLT2is and ARNIs are known to protect kidney function from deterioration, but do not improve it.25–27

There were no significant differences in LVEF between patients with and without events. Furthermore, patients who completely recovered LVEF (≥50% at discontinuation) did not have significantly fewer events than other patients; from our results, it is not possible to establish a link between LVEF and relapse of congestion. Other echocardiographic parameters (left ventricular global longitudinal strain, left atrial ejection fraction or left atrial volume index)28–30 could have a role in discriminating between patients with and without events, but this was not addressed in our study. Interestingly, patients with CRT devices at discontinuation had more events than those without, with significant differences. Despite CRT, these patients showed a trend for lower median LVEF, which could help explain this result.

A small number of previous studies have explored loop diuretic discontinuation in stable HF outpatients, with conflicting results. ReBIC-1, a double-blind, multicenter, randomized controlled trial (RCT), randomized 188 stable HF outpatients to either furosemide discontinuation or maintenance, with a follow-up of three months.9 The study reported that HF events were infrequent and did not differ between the furosemide discontinuation and maintenance groups. Furosemide resumption was also similar between groups, as was patient-reported perception of dyspnea, favoring furosemide discontinuation in stable outpatients with no signs of congestion. The findings of the present study are in line with those of the ReBIC-1 trial, as events were infrequent and few patients had to restart loop diuretics – but with the difference of a much longer follow-up period (one year in our study).

By contrast, another RCT that assessed loop diuretic omission in stable outpatients showed a clear diuretic response phase in chronic patients taking loop diuretics.8 In that trial, diuretic cessation caused a 50% drop in natriuresis and a 31% fall in urine output, underlining the risk of sodium and water retention, especially for patients with high N-terminal pro-BNP (NT-proBNP) such as those in that study. However, as pointed out in the study, while natriuresis has an important role in assessing acute HF diuretic response, it is not studied for diuretic discontinuation in chronic patients. Our study did not measure NT-proBNP but BNP, which was the NP available at our center at that time. Our patients had low median BNP at discontinuation, so the results cannot be extrapolated to patients with high NP levels. Results from the present study raise questions as to whether these changes in natriuresis and urine output seen in the acute phase really have a meaningful clinical impact in the chronic phase, as discontinuation of diuretics was possible without significant relapse of congestion.

Another RCT on diuretic discontinuation in elderly patients31 concluded that withdrawing diuretic therapy in HF would do more harm than good in most patients, as a significant number of patients in the withdrawal group needed to restart diuretics, as well as suffering a significant increase in blood pressure. However, in that trial, of the 102 patients in the withdrawal group, only 46 (45%) were prescribed diuretics with the specific indication of treating HF, with 42 patients taking diuretics to control hypertension, and only 31 patients (30.4%) were taking a loop diuretic (and combinations with other diuretics were included in this number), with most taking thiazides. The characteristics of HF patients in that trial were also not specified, and overall HF treatment was very different, as the article dates from 1997. Overall, it is possible that these results might not apply directly to loop diuretic discontinuation in euvolemic patients with HFREF.31

The study we conducted succeeded in collecting some novel information on a topic for which previous evidence is scarce, a strong point that can make it clinically and scientifically relevant. Following loop diuretic discontinuation patients for one year has not been achieved in the literature to date and is also noteworthy. The association we were able to establish between loop diuretic discontinuation and a small number of HF events in a particular subset of patients and conditions may be useful in clinical decision-making for similar patients and conditions. The data presented on patient characteristics and their clinical course until discontinuation and during follow-up, in a population in whom loop diuretic discontinuation could be achieved safely, may also be helpful when designing future studies on this topic.

This study has several limitations that should be considered, and its findings should be interpreted with caution. First and foremost, its observational and retrospective nature has inherent limitations, such as the availability of the desired information in EMRs, leading to missing data, which limits the interpretation of results. The collection of data from a single medical center is another potential source of bias. The absence of a control group is another clear limitation of our work. The small size of the sample considered also limits the statistical power to accurately compare patients with and without events and draw valid associations. Moreover, the study recorded inclusion appointments which occurred in different years, as did the subsequent discontinuation appointment and follow-up period, which impairs the comparability of the data. Notwithstanding, there were no major changes in HF treatment during the period of our study. Only three patients were taking torasemide, and thus our study cannot directly establish that its conclusions would be applicable to loop diuretics other than furosemide.

External validity is limited by the above patient characteristics. There is a need for more studies to better understand whether diuretic discontinuation would also be possible in different subsets of patients (such as older patients or those with impaired kidney function).

Loop diuretic discontinuation in a cohort of stable, euvolemic outpatients with HFREF was associated with a low number of episodes of diuretic resumption and WHF events, supporting the idea that loop diuretic discontinuation is possible in selected patients and conditions. Multicenter randomized controlled studies focused on loop diuretic discontinuation in euvolemic ambulatory patients, with larger sample sizes and prospective in nature, are needed to gather more conclusive evidence.

The authors have no conflicts of interest to declare.