Improvement in left ventricular ejection fraction after pharmacological up-titration in new-onset heart failure with reduced ejection fraction

We performed a retrospective longitudinal cohort study of consecutive patients with new-onset HF who presented to our tertiary care hospital between 2012 and 2018. Patients were included when they were referred to a specialised HF outpatient clinic, either after a first appointment with a cardiologist or 10 days after discharge following a first hospitalisation for HF. Patients had to be diagnosed with HF by a cardiologist no longer than 3 months before inclusion. In the outpatient clinic, GDMT was initiated and/or up-titrated by specialised HF nurses. Up-titration was done using pre-specified protocols based on the ESC guidelines for the diagnosis and treatment of HF (Electronic Supplementary Material, Table S1) [10, 11]. Furthermore, patients received education on self-care and had easy access to the HF clinic in the case of worsening HF. Clinical characteristics and doses of HF medication were recorded at the baseline and follow-up visits. We selected the visit that was performed 9 months after the baseline visit, or the visit that was closest to 9 months after baseline, to establish the success of up-titration and to assess changes in LVEF. We chose 9 months because patients were expected to be fully up-titrated by this time [12]. If patients had not achieved 100% of the guideline recommended target dose for either beta-blockers, ACEIs/ARBs/ARNIs or MRAs at this visit, the reason for incomplete up-titration was recorded. At the censor date of 1 September 2018, survival status and number of HF hospitalisations were recorded for every patient. This study complied with the Declaration of Helsinki and all national and local regulations as confirmed by the medical ethical evaluation committee. Patients were involved in the design and conduct of this research.

Echocardiography was performed as part of routine clinical care. A baseline echocardiogram was selected that was closest to the actual date of the baseline visit to the clinic. Patients were excluded if LVEF could not be reliably determined at baseline. A follow-up echocardiogram was selected closest to the 9‑months follow-up visit, and at least 6 months after the initial visit.

Normally distributed data are presented as mean (standard deviation). Data that were not normally distributed are presented as median (25th percentile, 75th percentile). Intergroup differences were tested using one-way ANOVA for normally distributed data. The chi-square or Kruskal-Wallis test was used for not normally distributed data. We performed linear regression to analyse determinants of change in ejection fraction. We considered any baseline demographic, clinical, laboratory and medication characteristic that was deemed important based on clinical reasoning for our multivariable analysis. Subsequently, stepwise backward selection was performed to derive the final model. Crude and multivariable adjusted Cox proportional hazard models were performed to evaluate the effect of changes in LVEF on mortality. All analysis were performed in R version 3.6.0.

In our cohort of new-onset HFrEF patients, the mean age of the 378 patients was 65.5 (±14.1) years and 34.1% were women. Baseline characteristics are presented in Tab. 1. A total of 345 patients had a follow-up visit, and 278 patients had echocardiography performed both at baseline and follow-up.

Use of beta-blockers and RAS inhibitors at any dose was already high at the baseline visit, with 90% of patients with HFrEF using any dose of beta-blockers. For RAS inhibitors, 87% used any dose at baseline. Use of MRAs was lower, with 41% of patients with HFrEF using an MRA at baseline (Tab. 1).

Of 345 HFrEF patients who were on medication, 69% reached ≥ 50% of the recommended dose of ACEIs/ARBs/ARNIs, 73% reached ≥ 50% of the recommended dose of beta-blockers, and 77% reached ≥ 50% of the recommended dose of MRAs (Tab. 2). Reasons for not achieving the target doses are presented in Tab. 3. For beta-blockers, the main reasons for not reaching the recommended target dose were hypotension (40.2%) and bradycardia (25.6%). For ACEIs/ARBs/ARNIs, the most important reason for not reaching the target dose was hypotension (54.0%), followed by renal dysfunction (14.7%) and hyperkalaemia (6.2%). The most important reason for not reaching the recommended dose of MRAs was that there was no longer an indication at the time of the follow-up visit because the patients’ functional status had improved to New York Heart Association (NYHA) functional class I. Patients who did not receive (the recommended dose of) a MRA at the time of the 9‑months visit, since there was no guidelines indication anymore, had a survival rate that was comparable to that of HFrEF patients who reached 100% of the target dose (3-year survival rate 88% vs 94% respectively). Patients with HFrEF who had a guideline indication for MRAs that received less than 100% of the target dose had considerably worse survival (79% at 3 years) than those on recommended doses.

During a median follow-up of 9 months, improvement from HFrEF to HF with mid-range EF/HF with preserved EF was seen in 131 of 344 of patients (35%). Fig. 1 shows a flow diagram of changes in LVEF categories. LVEF improved from a mean of 27.6% at baseline to 38.8% at follow up (+11.2%). Multivariable predictors of improvement of LVEF are presented in Tab. 4. Baseline LVEF was the strongest predictor of subsequent improvement. In addition, older age, non-ischaemic aetiology and higher plasma levels of N‑terminal pro-B-type natriuretic peptide were all significantly associated with an increase in LVEF in a multivariable linear model. A history of hypertension was associated with a decrease in LVEF.

For patients with HFrEF, 1‑year mortality was 10%, and mortality after 3.3 years was 22%. In patients with HFrEF, each 5% increase in LVEF was associated with a hazard ratio of 0.83 (0.72–0.96, p = 0.011) for mortality in a Cox proportional hazards model adjusted for age and sex, and a hazard ratio of 0.82 (0.73–0.93, p = 0.001) for the combined endpoint of mortality and/or HF hospitalisation (Tab. 5).

Several studies have investigated the effects of nurse-led disease management programmes in HF. Nurse-led care was associated with better patient-reported outcomes in a randomised study from Germany [8]. Planned referral to a nurse-led HF clinic was associated with lower risk of death (but not HF hospitalisation) in the Swedish Heart Failure Registry [9]. Similar results were found in the Dutch Deventer-Alkmaar study [14]. In contrast, in a multi-centre randomised trial (the Which Heart failure Intervention is most Cost-effective in reducing Hospital stay (WHICH? II) Trial) a structured, nurse-led, multidisciplinary management programme did not lead to a reduction in hospitalisation rate or all-cause mortality in Australia. There was, however, a better cardiac recovery on echocardiography (defined as improvement in either LVEF, left ventricular hypertrophy or E/e’) after 3 years of follow-up [15]. In the Coordinating Study Evaluating Outcomes of Advising and Counseling in Heart Failure (COACH) study with 1023 participants, investigating the effect of education and counselling in HF, there was no difference in all-cause mortality or HF hospitalisation after 18 months of follow-up [16]. In this latter study however, up-titration of HF medication was not part of the study protocol. The additive value of a nurse-led HF clinic might therefore depend on the healthcare system where it is initiated. Reasons why nurse-led HF clinics might be superior to standard care are the ability to closely monitor symptoms, optimise treatment by frequent dose adjustments, and the possibility of providing elaborate education on self-care and psychosocial support to patients with HF, including easy access to a healthcare provider in the case of deterioration.

In the present study, we carefully documented reasons for not reaching the recommended target doses in all patients. For beta-blockers, the most common reasons preventing further up-titration were hypotension, or associated complaints such as dizziness or light-headedness, and bradycardia. Hypotension is common in HF, and it can be difficult to distinguish disease-related hypotension from the effect of drugs. A recent meta-analysis found that individual beta-blockers did not exhibit a graded dose-response effect on systolic or diastolic blood pressure over the recommended dose range, suggesting that patients on lower doses might be able to tolerate higher doses as well [17]. In addition, there is evidence from a post hoc analysis of the COPERNICUS trial that patients with the lowest initial blood pressure had the highest improvement in quality of life if treated with carvedilol. In this group of HF patients, blood pressure increased instead of decreasing after treatment with a beta-blocker [18]. For ACEIs/ARBs/ARNIs, other frequent reasons hampering up-titration were renal dysfunction and hyperkalaemia. However, patients with incident hyperkalaemia who are maintained on ACEIs/ARBs/ARNIs might have better survival than those in whom ACEIs/ARBs/ARNIs were down-titrated or stopped because of hyperkalaemia [19].

The ESC HF guidelines provide a class IA recommendation to start with a beta-blocker and RAS inhibitor for all patients with HFrEF. An MRA is indicated in those patients who remain symptomatic despite treatment (NYHA class II or higher) and have a LVEF < 35%. In our cohort of patients with HFrEF, after 9 months of protocolised up-titration, a frequent reason for not receiving the target dose of MRAs was that patients had become asymptomatic (NYHA class I) or that their LVEF had improved. This raises the question whether introducing an MRA earlier, before target doses of beta-blockers and RAS inhibitors are achieved, might lead to improved outcomes. At present, there are few data available to answer this question.

We found a high percentage of improvement of LVEF in patients with new-onset HFrEF. This percentage is similar to that in a recent retrospective cohort study that found that in 38% of 3124 patients with HFrEF had an increase in LVEF of ≥ 10% after ≥ 6 months of follow-up [3]. In our cohort, patients with ischaemic heart disease were less likely to improve, whereas improvement was more likely in those with a dilated cardiomyopathy. Of note is that LVEF is subject to measurement variation. A study from 2012 concluded that around 20% of patients would be reclassified to a different category if two observers assessed the same echocardiogram and a single cut-off was used [20].