Long-Term Trajectories of Biomarkers, Functional, and Echocardiographic Parameters in Patients with Chronic Heart Failure from Dilated or Ischaemic Cardiomyopathy

Introduction: The long-term evolution of clinical, echocardiographic, and laboratory parameters of cardiac function in patients with chronic heart failure (HF) with either reduced (HFrEF) or mildly reduced (HFmrEF) left ventricular ejection fraction (LVEF) is incompletely characterised. Methods: We identified patients with chronic stable HF who presented at least twice to a university HF outpatient clinic between 1995 and 2021. Trajectories of NYHA functional class, LVEF, left ventricular internal end-diastolic diameter (LVIDD), NT-proBNP concentrations, and HF treatment over 10 years of follow-up were analysed using fractional polynomials. Analyses were repeated after stratifying patients according to aetiology (ischaemic vs. dilated) or HF category (HFrEF vs. HFmrEF). Results: A total of 2,132 patients were included, of whom 51% had ischaemic and 49% had dilated HF. Eighty six percent and 14% were classified as HFrEF and HFmrEF, respectively. Mean LVEF was 28 ± 10%, and median NT-proBNP and estimated glomerular filtration rate values were 1,170 (385–3,176) pmol/L and 81 (62–100) mL/min/1.73 m2, respectively. Median follow-up was 5.2 (2.6–9.2) years. Overall, NYHA functional class and LVIDD trajectories were U-shaped, whereas LVEF and NT-proBNP concentrations markedly improved during the first year and remained stable thereafter. However, the evolution of HF parameters significantly differed with respect to HF category and aetiology, with greater improvements seen in patients with HFrEF of non-ischaemic origin. Improvements in HF variables were associated with optimization of HF therapy, notably with initiation and up-titration of renin-angiotensin-system blockers. Conclusion: This study provides insights into the natural history of HF in a large cohort of well-treated chronic HF outpatients with respect to subgroups of HF and different aetiologists.


Introduction
Chronic heart failure (HF) is a global epidemic that affects about 64 million people worldwide and is increasing in prevalence [1].In developed countries, HF is a major cause for morbidity and mortality and a substantial driver of hospital admissions and resource utilization.Despite advances in HF treatment over the past 3 decades, 5-year mortality reaches up to 50-60% [1][2][3], and patients are on average hospitalized once every year after the initial HF diagnosis [4].
While mortality rates and predictors of prognosis have been studied extensively in HF [5][6][7][8][9][10], the longitudinal analysis of HF symptoms and parameters of cardiac function in surviving HF patients is incomplete.Knowledge about the natural history of HF, however, is crucial to inform patients, improve adherence, and guide decisions about further HF management.Prior reports have demonstrated an improvement in left ventricular ejection fraction (LVEF) in some patients with HF with reduced ejection fraction (HFrEF) in response to medical therapies [11][12][13][14][15][16][17].Yet, for these studies, follow-up was mostly short, the number of patients was limited and/or analyses were restricted to patients with full recovery of LVEF.In addition, the long-term trajectories of other HF variables such as N-terminal probrain natriuretic peptide (NT-proBNP) or symptom burden as well as the impact of HF category and aetiology are hitherto incompletely characterised.In the present study, we therefore present the long-term evolution of clinical, echocardiographic, and laboratory parameters of cardiac function in outpatients with chronic stable HF with respect to baseline HF category and aetiology.

Patient Selection and Follow-Up
Patients' data were extracted from the Heidelberg HF Registry as described below.All patients with stable chronic HF who attended the specialized HF outpatient clinic of the University Hospital Heidelberg, Germany, between 1996 and 2021 for evaluation of HF were offered inclusion into the local HF registry.Less than 1% refused to participate.Inclusion into the HF registry is continuous and on-going.To be eligible for this study, patients were selected from the registry according to the following inclusion criteria: 1. Diagnosis of chronic stable HF with depressed ejection fraction, 2. Individually optimized treatment with guideline-recommended HF therapies, 3. Signature of written informed consent for inclusion into the Heidelberg HF Registry, 4. At least two consecutive visits in our HF outpatient clinic, and 5. Documentation of the New York Heart Association (NYHA) functional class, LVEF, left ventricular internal end-diastolic diameter (LVIDD), and NT-proBNP at presentation to our HF outpatient clinic.The diagnosis of HF was established according to guidelines on the basis of typical symptoms and signs associated with an objective abnormality of cardiac structure or function on echocardiography, cardiac magnetic resonance imaging, or left heart catheterization [18].At baseline, all patients had a LVEF ≤49%, indicating HFrEF or HF with mildly reduced ejection fraction (HFmrEF).Stability of chronic HF was verified by the selection of outpatients who did not require any hospitalization for HF within 3 months prior to presentation to our HF outpatient clinic.Treatment was at the discretion of the treating physician with respect to guideline recommendations.

Patient Evaluation
NYHA functional class was assessed by a cardiologist taking into account the patient's HF symptoms and functional capacity as measured by 6-min walk test and/or cardiopulmonary exercise testing if available.Two-dimensional echocardiography was performed by image expert cardiologists and reviewed by expert staff.LVEF was obtained from apical 2-and 4-chamber views and calculated with the Simpson method.LVIDD was measured in end-diastole in parasternal long axis and apical 2-and 4-chamber views, timed with mitral valve closure at the level of the mitral valve chordae.
Blood samples were taken from a peripheral venous catheter using a lithium heparin vacutainer tube.Samples were centrifuged at 4°C immediately after collection to separate the plasma.NT-proBNP was measured according to the standard protocol of the fully automated Elecsys ® Roche Diagnostics analyser.NT-proBNP results are presented in pg/mL.To convert to pmol/L, results need to be multiplied by 0.118.
NYHA functional class, LVEF, LVIDD, and NT-proBNP were selected as variables of interest since they represent important measures of HF symptoms, cardiac function, cardiac structure, and volume status, respectively.HF treatment and dosing were recorded at each visit to the HF outpatient department.Target doses and dose equivalents for angiotensin-converting enzyme inhibitors (ACEI), angiotensin receptor blockers (ARB), and beta blockers were derived from current guidelines [10].For example, daily doses of 10 mg ramipril, 32 mg candesartan, or 50 mg carvedilol were considered as 100% dose equivalent, while 5 mg ramipril, 16 mg candesartan, or 25 mg carvedilol were defined as 50% dose equivalent.Loop diuretic dose was expressed in mg furosemide.Daily oral doses of 10 mg torasemide were considered to be equivalent to 40 mg furosemide.

Statistical Analysis
The data are presented as mean ± standard deviation, median (interquartile range), or number (%) as appropriate.To compare frequencies, χ 2 analysis was performed.To test for significant differences between groups, the Wilcoxon test and Student's t test were used where appropriate.All tests are two-tailed and an arbitrary p value of less than 5% was regarded as statistically significant.
Longitudinal analyses of clinical and functional HF variables were performed by use of fractional polynomials that were modelled using the STATA statistical software.Analyses included all available subsequent measurements of NYHA functional class, LVEF, LVIDD, or NT-proBNP for each patient.Analyses were repeated after stratification of patients according to HF category and aetiology of HF.Patients who underwent heart transplantation and those who died during follow-up were censored at the time of transplantation or death, respectively.

Baseline Characteristics and Follow-Up
We identified 2,132 patients who fulfilled the inclusion criteria outlined above.At baseline, mean age was 55 ± 15 years and 76% were male.The aetiology of HF was ischaemic in 51% and dilated in 49% of patients, and 86% and 14% of patients were classified as HFrEF and HFmrEF, respectively.Mean LVEF was 28 ± 10%, and median NT-proBNP and estimated glomerular filtration rate values were 1,170 (385-3,176) pmol/L and 81 (62-100) mL/min/1.73m 2 , respectively.Patients with ischaemic HF (ischaemic cardiomyopathy [ICM]) differed from those with dilated cardiomyopathy (DCM) in a number of variables (Table 1).Overall, patients with ICM were older, in a higher NYHA functional class and had higher baseline NT-proBNP levels than patients with DCM.In addition, patients in the ICM group had a higher burden of cardiovascular comorbidities including arterial hypertension, diabetes, and chronic obstructive lung disease.Patients with HFmrEF at baseline were more often female, had less HF symptoms and lower NT-proBNP levels than those with HFrEF.In addition, fewer patients were treated with loop diuretics and mineralocorticoid receptor antagonists in the HFmrEF group.Baseline characteristics of patients with ICM and DCM with respect to HF category are presented in online supplementary eTables 1 and 2 (for all online suppl.material, see https://doi.org/10.1159/000532070).Median follow-up of surviving patients was 5.2 (2.6-9.2) years.

Evolution of NYHA Functional Class
A total of 12,552 NYHA functional class assessments were performed during follow-up, corresponding to a mean of 5.9 assessments per patient.At baseline, mean NYHA functional class was 2.1, with 565 (27%), 704 (34%), 781 (37%), and 26 (1%) patients being classified as NYHA I, II, III, and IV, respectively.At follow-up, mean NYHA functional class was 2.2, and the corresponding numbers for patients in NYHA functional classes I, II, III, and IV were 651 (32%), 587 (28%), 747 (35%), and 52 (2%).At the time of the last available NYHA assessment, 557 (26%) patients had experienced an improvement and 425 (20%) a worsening of their NYHA functional class.Figure 1a presents NYHA functional class trajectories in the entire study cohort.In the whole patient sample, the NYHA functional class trajectory was faintly U-shaped, with a slight improvement during the first 3 years of follow-up, a plateau from years three to eight and a worsening thereafter.After 10 years of follow-up, mean NYHA functional class was similar to that at baseline.Patients with DCM showed greater improvements in NYHA functional class over time than those with ICM (30% vs. 23% with at least 1 class improvement at last available follow-up as compared to baseline, p < 0.001).In addition, a worsening in NYHA functional class was less frequent in DCM than in ICM patients (17% vs. 23%, p < 0.001).As shown in Figure 1b, the NYHA functional class trajectory in patients with DCM was clearly U-shaped, with a constant improvement during the first 6 years and a worsening starting at 8 years of follow-up.In contrast, some patients with ICM experienced a rapid improvement of HF symptoms shortly after the initial presentation to our HF outpatient department without any subsequent changes in mean NYHA functional class during further follow-up.
The proportion of patients with an improvement of at least one NYHA functional class at last available followup tended to be higher in patients with HFrEF as compared to those with HFmrEF (27% vs. 22%, p = 0.07, Fig. 1c).NYHA functional class trajectories stratified by aetiology and HF category are shown in online supplementary eFigure 1A-D.Changes in NYHA class between the first and last available measurement correlated with changes in LVEF and NT-proBNP irrespective of the aetiology of HF.However, NYHA class was related to LVIDD only in patients with DCM but not in those with ICM (online suppl.eTable 3).

Evolution of LVEF
A total of 12,637 LVEF measurements were performed during follow-up, corresponding to a mean of 5.9 assessments per patient.Mean LVEF at baseline was 28 ± 10%, with no significant differences between ICM and DCM patients (p = 0.11).At last available follow-up, LVEF had increased by 3 ± 0.10% to 31 ± 13%.An improvement in LVEF of ≥5% was noticed in 733 patients (34%), whereas a decrease of ≥5% was observed in 371 patients (17%) (Table 2).As shown in Figure 2a, there was a rapid improvement of LVEF within the first year of follow-up followed by a plateau up to a decade.The overall improvement of LVEF was greater in patients with DCM than in those with ICM (Fig. 2b).Significantly more patients with DCM showed an increase in LVEF of ≥5% at last available follow-up (41% of patients with DCM vs. 28% of patients with ICM, p < 0.001), and a decrease in LVEF of ≥5% was less frequent among patients with DCM (14% of patients with DCM vs. 21% of patients with ICM, p < 0.001).The proportion of patients with a significant improvement in LVEF at last available followup was similar in HFrEF and HFmrEF (35% vs. 32%, p = 0.35).However, LVEF trajectories significantly differed with respect to HF category (Fig. 2c).Among patients with HFrEF at baseline, 251 (14%) improved to HFmrEF at study end, whereas 62 (20%) of HFmrEF patients deteriorated to the HFrEF category.Overall, 52% of HFmrEF patients changed their HF category over time as opposed to 14% of HFrEF patients.In multivariable regression analyses, younger age, female sex, non-ischaemic HF, higher LVEF, and absence of treatment with anticoagulants and loop diuretics were significantly associated with the transition towards a better HF category in patients with HFmrEF at baseline (online suppl.eTable 4).In contrast, LVEF and LVIDD were independently associated with the transition from the HFmrEF to the HFrEF category during follow-up (online suppl.eTable 5).LVEF trajectories stratified by aetiology and HF category are shown in online supplementary eFigure 2A-D.As shown in online supplementary eFigure 3, LVEF trajectories were similar in patients with mild versus severe HF symptoms.However, mean LVEF was higher in patients with NYHA functional class I or II symptoms as compared to those in class III or IV.

Evolution of LVIDD
A total of 10,136 LVIDD measurements were performed during follow-up, corresponding to a mean of 4.8 assessments per patient.Mean LVIDD was 60 ± 10 mm at baseline and 59 ± 10 mm at follow-up.At last available follow-up, LVIDD had increased ≥10% in 369 (17%) and decreased ≥10% in 406 (19%) patients, respectively (Table 2).As shown in Figure 3a, the course of LVIDD was slightly U-shaped, with a decrease during the first 3 years, a plateau between years three and eight and a reincrease thereafter.At baseline, LVIDD was higher in patients with DCM than in those with ICM (61 ± 11 mm vs. 58 ± 10 mm, p < 0.001).However, no significant difference in LVIDD between DCM and ICM patients was noted at study end (59 ± 11 mm vs. 59 ± 10 mm, p = 0.67).As presented in Figure 3b, LVIDD steadily decreased in patients with DCM, whereas it steadily increased in patients with ICM.LVIDD was significantly greater in patients with HFrEF than in those with HFmrEF irrespective from HF aetiology (Table 1; online suppl.eTables 1 and 2).In patients with HFrEF, the course of LVIDD was slightly U-shaped, whereas LVIDD steadily increased in the HFmrEF group (Fig. 3c).LVIDD trajectories stratified by aetiology and HF category are shown in online supplementary eFigure 4A-D.Changes in LVIDD between the first and last available measurement significantly correlated with changes in NT-proBNP (r = 0.06, p = 0.005 and r = 0.07, p < 0.001 for absolute and relative changes, respectively).However, this was only true in patients with DCM (online suppl.eTable 6).

Evolution of NT-proBNP
A total of 10,397 NT-proBNP measurements were performed during follow-up, corresponding to a mean of 4.9 assessments per patient.Median NT-proBNP concentration was 1,170 (385-3,176) pg/mL at baseline and 860 (211-3,515) pg/mL at study end, with higher baseline levels in patients with ICM as compared to DCM (Table 1).At last available follow-up, median NT-proBNP values were unchanged in ICM patients but had halved in those with DCM when compared to baseline (1,353 [454-4,192]   to baseline, NT-proBNP had increased ≥30% in 634 (30%) and decreased ≥30% in 665 (31%) patients, respectively (Table 2).There was a substantial decrease in NT-proBNP concentrations shortly after inclusion in our HF registry and NT-proBNP values remained at lower levels thereafter (Fig. 4a).The proportion of patients with a substantial decrease in NT-proBNP values at study end was higher in the DCM as compared to the ICM group (37% vs. 16%, p < 0.001).Similarly, fewer patients with DCM experienced a significant increase in NT-proBNP (24% with DCM vs. 36% with ICM, p < 0.001).In both patients with DCM and ICM, changes in NT-proBNP values occurred early after inclusion into our HF registry (Fig. 4b).As shown in Figure 4c, the early drop in

Discussion
In the present longitudinal long-term study of 2,132 patients with chronic stable HF with reduced systolic function, we observed a significant improvement of LVEF and NT-proBNP concentrations at 1 year followed by a plateau for up to a decade.In contrast, NYHA functional class and LVIDD trajectories were U-shaped, with slight improvements noted over the first 4 years of follow-up.However, the evolution of HF variables was highly variable with respect to HF category and aetiology, with most favourable trajectories being observed in patients with HFrEF from non-ischaemic origin.Improvements in HF variables were associated with optimization of HF treatment, notably with initiation and up-titration of renin-angiotensin-system blockers.
Our data provide insight into the natural history of HF in a large, well-characterised real-life cohort of outpatients with chronic stable HF and preserved renal function.Our study complements and expands on prior observational studies by providing long-term longitudinal data on a range of important HF variables in a representative chronic HF population.While most studies only report single assessments of parameters of cardiac function obtained at baseline, changes in HF variables and treatments are common and may affect morbidity and mortality of HF patients [12,13,[18][19][20][21].Therefore, an understanding about the natural history of chronic HF would help inform patients, improve adherence, and guide decisions about the frequency of follow-up, need for repeat cardiac imaging and biomarker testing, and prognostic counselling.
Our study confirms and expands prior observations on LVEF trajectories in patients with HF.In a longitudinal study of 1,160 HF patients with initially depressed LVEF, Lupón et al. [16] found an inverted U-shaped LVEF trajectory with a ≈8% rise in LVEF during the first year, a plateau up to a decade, and a slow decline thereafter.Similar to our data, this pattern was more pronounced in HF of non-ischaemic origin and in patients with HFrEF.In addition, Martinsson et al. [17] reported that LVEF improved on average by 9.1% in patients with HFrEF and 4% in patients with HFmrEF over 1 year in 201 patients enrolled in a contemporary outpatient-based HF management program in Sweden.Again, improvements were greater in patients with non-ischaemic heart disease.Similar results were reported from a community cohort study of 1,233 incident HF patients [13].
The early improvement in LVEF observed in our and the aforementioned studies likely reflects the positive effects of an optimized HF treatment on cardiac function.
In the present study, the prescription rates and dosing of ACEI/ARB significantly increased in parallel with LVEF over the first year of follow-up.Up-titration of diseasemodifying treatments was more common in patients with DCM, which may contribute to the more favourable LVEF trajectories in patients with non-ischaemic HF.In addition, LVEF recovery in ischaemic HF may be limited by non-contractile cardiac scar tissue, whereas nonischaemic HF patients show a pattern of temporal myocardial remission in response to treatment.This observation is supported by a recent study of 1,546 HF patients following sacubitril/valsartan treatment for 8 months, where patients with DCM had a higher degree of LVEF improvement than those with ICM [22].The initial increase in LVEF observed in our study, however, was somewhat smaller than that reported from previous studies [16,17].This may be explained by the inclusion of patients with chronic rather than new-onset HF, since the majority of patients had already started HF treatment before presenting to our referral HF outpatient clinic.
In the present study, half of the patients initially classified as HFmrEF changed their HF category during follow-up.Similarly, data from of the Swedish HF Registry show that 62% of patients with HFmrEF transitioned to a different HF category during 1.4-year follow-up [21].In addition, Lupón et al. [16] reported that only one-third of patients initially classified as having HFmrEF remained in that category during the full-term 15-year follow-up, whereas one-third evolved to HFpEF and one-fourth were categorized as having HFrEF.However, HFmrEF covers only a small range of LVEF and patients with HFmrEF are therefore more likely to change HF category than patients with HFrEF.Whether HFmrEF represents a distinct pathophysiological entity or a transitional phenotype between HFrEF and HFpEF is the subject of debate [23].Our long-term data support the characterization of HFmrEF as just a snapshot on the way towards recovering or declining LVEF rather than as a stable phenotype.
In parallel with the initial rise in LVEF, we observed a significant decrease in NT-proBNP values that levelled off approximately 1 year after study inclusion.This was especially true for patients with non-ischaemic HFrEF.Our data are consistent with prior studies that showed a decline in NT-proBNP concentrations in response to HF treatment in the short term [17,24,25].However, to the best of our knowledge, our study is the first to present long-term trajectories for NT-proBNP in a cohort of stable outpatients with chronic HF.
Despite the marked early improvements of LVEF and NT-proBNP, we observed only a small reduction in LVIDD during the first years of follow-up.This reduction was evident in DCM and HFrEF patients only, whereas LVIDD steadily increased in patients with HFmrEF or ICM.In addition, changes in LVIDD over time correlated with changes in NT-proBNP in patients with DCM but not ICM.In a cohort study of 168 consecutive HF patients followed for 40 ± 19 months, the magnitude of left ventricular size reduction at study end was significantly greater in DCM as compared to ICM patients [26].A reduction in LVIDD over time may indicate decongestion and/or reverse remodelling in response to HF treatment.Studies have shown a reduction of LVIDD in response to treatment with sacubitril/valsartan, dapagliflozin, eplerenone, or valsartan for up to 22 months [27][28][29][30][31].Given the differences in LVIDD reduction in DCM versus ICM, it is possible that current optimal medical therapy produces more favourable ventricular remodelling in DCM patients.
Although baseline LVEF was similar in ICM and DCM groups, patients with ICM suffered from more severe HF symptoms throughout the entire follow-up period.This observation is supported by another cohort study [26].However, to best of our knowledge, our study is the first to present long-term NYHA trajectories in patients with chronic HF.The higher symptom burden in patients with ICM may result from a higher age and more comorbidities as compared with the DCM group.In addition, we observed constantly higher NT-proBNP concentrations in ICM versus DCM patients.Natriuretic peptide release may result from increased cardiac filling pressures which have been associated with dyspnoea and thus a higher NYHA functional class.The rapid early symptom relief observed in some ICM patients may result from decongestion by optimization of diuretic treatment.Accordingly, we observed an adjustment of loop diuretic dosing over the first months of follow-up.

Limitations
This is a retrospective analysis from a prospective HF registry.The intrinsic limitations of any post hoc analysis should therefore be considered.We acknowledge that there is an intra-observer and inter-observer variability of LVEF, LVIDD, and NYHA assessments; however, considering the large number of assessments performed, we assume that such variability was randomly distributed during follow-up [32][33][34][35][36][37].LVEF and LVIDD were assessed by transthoracic echocardiography in routine clinical care.Three-dimensional echocardiography or cardiac magnetic resonance imaging would evaluate left ventricular function and diameters more precisely, but they are not broadly available in clinical practice.We present the evolution of LVEF and LVIDD measurements as these variables are commonly used and broadly available markers of cardiac structure and function.However, the explanatory value of LVIDD may be limited in patients with ICM who may present with asymmetric dilations of the left ventricle due to segmental alterations of contractility.Also, we do not have data on revascularisation therapies in patients with ischaemic HF during follow-up, which may impact parameters of cardiac function as well as HF symptoms.HF symptom burden was assessed by the referring physician considering the patient's history, clinical examination, and functional capacity.However, studies have shown poor agreement between patient-and provider-exhibited NYHA assignments [38].Validated questionnaires such as the Edmonton Symptom Assessment Scale (ESAS) or the Kansas City Cardiomyopathy Questionnaire (KCCQ) would have allowed a more comprehensive assessment of the HF symptoms, but unfortunately, they were not available in our study.Unfortunately, we do not have data on the index HF variables at first diagnosis of HF, since the majority of patients are referred to our outpatient clinic by resident cardiologists or general physicians who have already started HF treatment before referral.Also, we cannot comment on mortality or hospitalization rates of patients included in the present study.This is because we only included patients who presented at least twice to our HF outpatient clinic.In survival analyses, however, the a priori exclusion of all patients who died after their first visit to our clinic would cause considerable bias.In addition, literature already provides robust evidence on the prognostic value of HF variables such as LVEF or NT-proBNP [10,12,13,18,20,21].The selection of surviving patients may also have contributed to the relatively high number of patients with no or mild HF symptoms in our study.The present study included patients over 25 years who were referred to a specialized university HF clinic.During that time, the definition of HF varied and it was not until 2021 that the universal definition of HF was published [39].Patient selection for inclusion into the Heidelberg HF Registry may thus have changed over time.However, the definition of HF has always relied on symptoms and signs of HF in addition to evidence of cardiac dysfunction [40].As the definition of HF particularly changed with respect to HF with preserved LVEF, we restricted our analyses to patients with evidence of depressed LVEF in order to prevent selection bias.Also, it is important to note that our study cohort comprised ambulatory patients with chronic stable HF and preserved renal function.Results may not be transferable to other study cohorts such as patients with de novo HF or concomitant kidney failure.Over the past decades, HF treatment has evolved significantly, with four new classes of drugs (angiotensin receptor neprilysin inhibitors, sodium-glucose 2 cotransporter inhibitors, myosin activators, and guanylate cyclase stimulators) being approved to market in recent years.Future studies should investigate the impact of new HF therapies on the long-term evolution of HF variables in patients with chronic HF.

Conclusion
This retrospective observational study provides insights into the natural history of HF in a large cohort of welltreated chronic HF outpatients.We found a marked improvement of LVEF and NT-proBNP concentrations at 1 year followed by a plateau for up to a decade.In contrast, NYHA functional class and LVIDD trajectories were U-shaped, with slight improvements noted over the first 4 years of follow-up.However, the evolution of HF variables was highly variable with respect to HF category and aetiology, with most favourable trajectories being observed in patients with HFrEF from non-ischaemic origin.Improvements in HF variables were associated with optimization of HF treatment, notably with initiation and uptitration of renin-angiotensin-system blockers.

Fig. 1 .
Fig. 1.Long-term evolution of mean NYHA functional class in patients with chronic HF in the entire study cohort (a), stratified by aetiology of HF (b), and stratified by HF category (c).

Fig. 2 .
Fig. 2. Long-term evolution of LVEF in patients with chronic HF in the entire study cohort (a), stratified by aetiology of HF (b), and stratified by HF category (c).

Fig. 3 .
Fig. 3. Long-term evolution of LVIDD in patients with chronic HF in the entire study cohort (a), stratified by aetiology of HF (b), and stratified by HF category (c).

Fig. 4 .Fig. 5 .
Fig. 4. Long-term evolution of NT-proBNP in patients with chronic HF in the entire study cohort (a), stratified by aetiology of HF (b), and stratified by HF category (c).

Table 2 .
Changes of important HF variable between baseline assessment and last available follow-up Significant p values are written in italics.HF, heart failure; ICM, ischaemic cardiomyopathy; DCM, dilated cardiomyopathy; HFrEF, heart failure with reduced ejection fraction; HFmrEF, heart failure with mildly reduced ejection fraction; NYHA, New York Heart Association functional class; LVEF, left ventricular ejection fraction; LVIDD, left ventricular internal end-diastolic diameter; NT-proBNP, N-terminal probrain natriuretic peptide; ACEI, angiotensin-converting enzyme inhibitor; ARB, angiotensin receptor blocker.