Is diuretic withdrawal safe in patients with heart failure and reduced ejection fraction? A retrospective analysis of our outpatient cohort

Although diuretics are essential to optimize volume status in patients with heart failure with reduced ejection fraction (HFrEF) [ ], safety and benefits of prolonged diuretic treatment are uncertain, in particular on patients with chronic and compensated HFrEF. In facts, diuretics can increase the neuro-humoral activation in HFrEF [ ], determine electrolyte disturbances [ ] and acute renal insufficiency [ ]. On the contrary, Hopper et al. identified seven studies of diuretic withdrawal (DW) in stable chronic HF, in which clinical decompensation was more frequent in the DW group [ ]. Our hypothesis was that diuretic therapy could be safely suspended in patients with HFrEF after adequate therapeutic neuro-hormonal antagonism [ ]. Therefore, we retrospectively analyzed our cohort of HFrEF outpatients with a twofold aim: first, to asses if DW is safe in patients with HFrEF and second, to identify clinical, biochemical or echocardiographic parameters associated to DW among HFrEF patients regularly followed-up at our outpatient's clinic. The study was conducted in our tertiary level HF clinical center. All ambulatory patients referred from November 2011 through September 2014 at our center were considered for recruitment. Exclusion criteria were as follows: (I) hospitalization for HF within 30 days before ambulatory evaluation, (II) myocardial revascularization and/or resynchronization therapy within 180 days before ambulatory visit, (III) congenital heart disease, (IV) severe valve heart disease. Overall, 216 consecutive clinically stable HFrEF patients (ejection fraction HF ≤ 35%) were evaluated for retrospective enrollment. Among them, 26 patients were excluded, since they were not taking diuretics. Eventually, 190 patients were recruited and stratified according to diuretics continuation: no withdrawal group (NWG), N = 169 (89%), and diuretics withdrawal group (WG) N = 21 (11%) (Fig. 1S in supplementary materials). Furosemide was the only diuretic used. Patients were assessed at baseline visit. Medical history, physical exam, 12-lead electrocardiogram and laboratory analysis comprehensive of N-terminal pro brain natriuretic peptides (NT-proBNP) and highly sensitive troponin essay were obtained. To assess the safety of DW, furosemide dose in WG was tapered down (steps of 25 mg) at each visit usually at the sixth month re-evaluation, according to the following criteria: 1) no symptoms or signs of congestion were evident at the clinical assessment and 2) NT-proBNP trend was in downturn. A standard 2-Dimensional and Doppler transthoracic echocardiogram was performed at baseline visit and repeated in case of worsening clinical conditions. Baseline characteristics between WG and NWG were compared using Mann-Whitney U test and Fisher exact test. Changes from baseline characteristics at follow-up were tested by paired t-or McNemar test. Univariate Logistic regression was used to identify baseline characteristics associated to diuretic withdrawal, while ROC analysis identified best cut-off values to differentiate WG and NWG groups. Two models of multivariate analysis were built: one including baseline echocardiographic parameters and the other including NT-proBNP and TAPSE. Furosemide dosage at baseline was forced in both models using SAS JMP 9 software package. NWG patients were older, had higher body mass index (BMI), had more frequently history of arterial hypertension, atrial fibrillation and were more commonly cardiac resynchronization therapy and implantable cardioverter defibrillator (ICD) carriers. No differences were found between the two groups concerning medical treatment except for diuretics. According to univariate logistic regression analysis, younger age, higher BMI, history of arterial hypertension, the presence of ICD and diuretic dose were associated with lower chance of discontinuing diuretics (Table 1). Patients in WG had lower NT-proBNP levels (Table 2). According to univariate regression analysis: NT-proBNP < 550 pg/ml and estimated glomerular filtration rate were associated with DW. Regarding echocardiography findings a LVEDD > 60 mm, TAPSE ≥ 20 mm were associated to DW (Table 2).