Impact of iron deficiency on congestion and postdischarge survival in patients recently hospitalised for decompensated heart failure: a multicentre, prospective, observational analysis of the FERIC-RO study

Introduction

Worsening chronic heart failure (WCHF) is a common cause of urgent hospitalisations in adults.1 The 1-year mortality following heart failure (HF) hospitalisation ranges between 25% and 30%, with more than 45% combined deaths and rehospitalisations.2–4

In patients hospitalised for acute HF (AHF), comorbidities are highly prevalent irrespective of left ventricular ejection fraction (LVEF) and associated with worse postdischarge outcomes,5–7 and represent barriers to implementation of HF therapies.8 Iron deficiency (ID) is one of the most commonly reported comorbidities in HF and is associated with higher rates of hospitalisations, worse functional status, poor quality of life and limited exercise capacity irrespective of the presence of baseline anaemia.9 10

For patients hospitalised with AHF, in-hospital management relies on decongestion and initiation or optimisation of guideline directed medical therapies (GDMTs) before discharge.4 In addition to clinical residual congestion, which portends a negative prognosis, a significant proportion of patients without clinical signs of congestion may still have high NT-proBNP values or insufficient NT-proBNP decrease, suggesting the presence of haemodynamic congestion, which is also associated with poor prognosis.11 To note, some comorbidities, such as diabetes, renal dysfunction and anaemia, were independently associated with persistent residual congestion in patients hospitalised for AHF.2 Although pathophysiologically plausible, currently, there is little data to show that ID may be associated with residual congestion, either clinical or haemodynamic, in patients hospitalised for AHF. In the current manuscript, we comprehensively described the prevalence and factors associated with ID, in patients hospitalised for AHF across the entire LVEF spectrum. Furthermore, we evaluate if ID may be associated with inadequate decongestion based on the clinical signs and symptoms at discharge, aiming to confirm this hypothesis using objective parameters, such as in-hospital decrease of NT-proBNP.

Methods

FiER deficit in Insuficienta Cardiaca in ROmania (FERIC-RO) registry was a prospective, observational study, conducted in nine academic hospitals in Romania. The design and methodology of the study were previously published.12 Patients were included between March 2018 and June 2021, allowing subinvestigators to enrol patients for a maximum period of up to 8 months. Centres enrolled patients at different time points within the specified period, resulting in a longer overall duration of enrollment. During recruitment period, all consecutive patients with HF were enrolled in the study at each site. Briefly, the study included adult patients with a history of HF hospitalisation in the last year before enrollment, with signs and symptoms of HF, and increased levels of natriuretic peptides (NPs) (ie, NT-proBNP>1000 pg/mL) who required hospital admission and were discharged alive. Patients with de novo AHF, those with cardiogenic shock, acute coronary syndromes in the last 30 days, patients already on iron therapy and those with a life expectancy of less than 6 months were excluded.12

Clinical congestion assessment, including dyspnoea, lower limb oedema, altered mental status, pulmonary oedema, pulmonary rales and jugular vein distension, was performed at admission and discharge. Echocardiography was performed in the first 24 hours of hospitalisation. At discharge, ID was screened in all patients using ferritin, transferrin saturation (TSAT) and soluble transferrin receptors (sTfRs), and defined as absolute when ferritin<100 ng/mL and functional when ferritin was between 100 and 299 ng/mL, and TSAT<20%. Study patients were divided into two groups: ID+ and ID− based on the international clinical practice guidelines definition.4 At 30 days after discharge, full blood count and iron profile (ferritin, TSAT and sTfRs) were repeated. Anaemia was defined as haemoglobin levels<13 g/dl in men or<12 g/dl in women. NT-proBNP was measured at admission and at discharge and the relative change between admission and discharge was reported.

The study was approved by the local ethics committee, according to the guidelines of the 1975 Declaration of Helsinki, and written informed consent was obtained from each patient. FERIC-RO study had three study visits: during the index hospitalisation, an on-site 30 days after discharge and 90-day telephone follow-up to assess vital status and rehospitalisations for HF.

The study had an international adjudication committee (AA, SC, PSP and JB) to confirm HF as a cause of hospitalisation. The data are not publicly available currently as the existing statistical reports are still ongoing.

Results

Variables associated with ID

In the univariate regression model, predictors of ID were anaemia (OR 3.454, 95% CI 1.810 to 6.593, p<0.001) and CKD (OR 2.060, 95% 1.053 to 4.028, p 0.03). Otherwise, ID was present irrespective of gender—males (OR 0.741, 95% CI 0.380 to 1.445, p 0.379), LVEF<40% (OR 0.977, 95% CI 0.524 to 1.819, p 0.941), ischaemic aetiology (OR 1.228, 95% CI 0.641 to 2.351, p 0.536) or diabetes mellitus (OR 0.684, 95% CI 0.367 to 1.276, p 0.233) (figure 1).

Figure 1

Request permissions

Variables associated with ID presented in forest plot graph. ID, iron deficiency; LVEF, left ventricular ejection fraction.

Congestion assessment

There were differences in the clinical signs of congestion at admission. Patients with ID+ had more rales (51.7% vs 41.9%, p 0.212), more peripheral oedema (75.3% vs 58.1%, p 0.020), jugular vein distension (65.2% vs 45.9%, p 0.014), dyspnoea at rest (73.0% vs 55.4%, p 0.019) and acute pulmonary oedema (7.9% vs 2.7%, p 0.151) compared with ID− (table 1,figure 2). Patients with ID+ had lower respiratory variation of inferior vena cava at transthoracic echocardiography than ID− (OR 0.340, 95% CI 0.166 to 0.695, p 0.003). At discharge, the proportion of patients with New York Heart Association (NYHA) III and IV was significantly higher in ID+ compared to patients with ID− (58.4% vs 31.1%; p 0.002) (table 1).

Figure 2

Request permissions

Clinical signs of congestion in the two groups at admission. ID, iron deficiency.

In patients with ID+, mean NT-proBNP levels at admission were 9288.3±7455.4 pg/dL compared with 4414.0±4843.9 pg/dL in ID− (p<0.001) (table 1). NT-proBNP levels higher than 1000 pg/dL at discharge, were more frequent in ID+ (80.2% vs 26.9%, p<0.001) and were associated with ID (OR 0.090, 95% CI 0.042 to 0.195, p<0.001). Average NT-proBNP percentage decrease of −45.7% was reported in ID+ versus −63.3% in ID− population (p 0.003), with higher absolute NT-proBNP values at discharge in ID+ (figure 3). Patients ID+ required, on average, an extended duration of intravenous therapies by 0.4 days (days with intravenous therapy in ID+ 5.4 days vs ID− 5.0 days, p 0.389), with no differences in LOS (ID+ 7.3 days, ID− 7.9 days, p 0.285).

Figure 3

Request permissions

Percentage change of NT-proBNP between two groups at discharge—graphical representation of percentage change during hospitalisation. ID, iron deficiency.

Clinical outcomes

At the first follow-up visit, all-cause mortality was reported in five patients with ID+ and two patients with ID− (5.6% vs 2.7%, p 0.361). At 90-days postdischarge, an additional 16 deaths occurred in ID+ versus 3 deaths in the ID− group (total 90-day mortality, 30.9% ID+ and 9.6% ID−, p 0.005). There were numerically more HF rehospitalisations in patients with ID+ at 30 days that did not reach statistical significance (27.8% vs 19.6%, p 0.338). At 90 days, patients with ID+ experienced 32 rehospitalisations compared with 12 rehospitalisations patients with ID− (47.1% vs 23.1%, p 0.007) (figure 4). Overall, at 90 days, patients with ID+ had more events (combined deaths and HF rehospitalisations) compared topatients with ID− (49 events—72.1% vs 17 events—32.7%, p<0.001).

Figure 4

Request permissions

Rates of mortality and HF rehospitalisations at 30 and 90 days between the two groups. HF, heart failure; ID, iron deficiency.

In a binary regression model, ID (OR 5.310, 95% CI 2.422 to 11.641, p<0.001) and anaemia (OR 3.178, 95% CI 1.502 to 6.724, p 0.002) were correlated with clinical outcomes (deaths and rehospitalisations) at 90 days. Also, a correlation of iron parameters on clinical outcomes was assessed, significant for TSAT (OR 0.929, 95% CI 0.889 to 0.972, p 0.001) and sTfR (OR 1.774, 95% CI 1.234 to 2.551, p 0.002), without significant impact of the ferritin (OR 0.999, 95% CI 0.998 to 1.001, p 0.509) (figure 5).

Figure 5

Request permissions

Association of iron parameters, ID and anaemia with combined clinical outcomes (deaths and HF rehospitalisations) at 90 days. HF, heart failure; ID, iron deficiency; sTfR, soluble transferrin receptor; TSAT, transferrin saturation.

In addition, a Cox regression model of 3-month rehospitalisations was performed (table 2). Through the analysed cofounders, hypertensive aetiology (HR 4.645, 95% CI 1.221 to 17.676, p 0.024), heart rate at discharge (HR 1.036, 95% CI 1.011 to 1.061, p 0.004) and ID (HR 2.165, 95% CI 1.029 to 4.555, p 0.042) were associated with the 3-month rehospitalisations rate.

Table 2

View inline•Open as popup

Multiple Cox regression model of 3-month hospitalisations

Discussions

The FERIC-RO registry enrolled a population of patients hospitalised for WCHF, with any LVEF and various aetiologies, who were screened for ID at the end of hospitalisation. The prevalence of ID was 54.6%, with a larger contribution of the functional ID, and there was no significant difference in ID’s prevalence between discharge and 1 month (54.6% vs 51.3%, p 0.782). Similar values of ID prevalence were reported when patients with WCHF were classified by ejection fraction. ID was associated with more clinical signs of congestion and higher NT-proBNP values at admission. At discharge, compared with ID−, patients with ID+ had more severe NYHA class, higher NT-proBNP values and lower in-hospital decrease of NT-proBNP. At 90-day postdischarge, patients with ID+ had significantly more deaths and HF rehospitalisations, compared to patients with ID−, persistent after adjusted for cofounders (table 2).

Several other studies that previously characterised ID in patients with WCHF have reported similar prevalences to our results (54% in HFrEF and 56% in HFpEF)13 or a significantly higher prevalence (69% in men and 75% females).14 These findings highlight that ID is a highly prevalent comorbidity in HF, irrespective of LVEF or biological sex.

In FERIC-RO, 55.1% of patients with ID+ had functional ID. In functional ID, iron is sequestered in the reticuloendothelial system, secondary to HF-induced proinflammatory status associated with long-term evolution of HF. This is a common finding in WCHF as compared with de novo HF15 and may explain the higher prevalence of functional ID as compared with previous studies enrolling any phenotype of AHF.16 FERIC-RO inclusion criteria allowed only patients with a history of HF, thus excluding patients with de novo HF.

TSAT, as a marker of the quantity of circulating iron, remains a major diagnostic tool for ID in these patients since ferritin values are not linearly associated with iron availability and may be influenced by other confounders, particularly infection and inflammation. Of ID parameters, only TSAT was significantly associated in binary regression model with 90-day clinical outcomes, while ferritin had no association. One of the strengths of our study is the collection of sTfRs that are highly expressed in low iron body stores and are independent of inflammation. In our univariate analysis, sRT was associated with 90-days outcome (all-cause mortality and HF rehospitalisations) (OR 1.774, 95% CI 1.234 to 2.551, p 0.002).

Currently, there are no recommendations for regular measurements, although previous studies reporting a correlation of high levels of soluble receptors for transferrin and mortality17 proposed newer definition for ID using hepcidine and sTfR. The current findings support a more comprehensive approach of the laboratory tests used to diagnose ID in AHF, particularly in WCHF settings associated with persistent low-grade systemic inflammation. Further research considering a new definition of ID, including parameters associated with postdischarge outcomes, may help to clarify the relationship between intravenous iron therapy and clinical endpoints.

Demographical, clinical and biological variables associated with ID may vary according to the methodology of the research protocols, but typically these relate to female sex, advanced heart disease, high levels of NPs and markers of inflammation.18 In FERIC-RO, anaemia and CKD were identified as independent predictors of ID. CKD, as a proinflammatory systemic condition, is highly prevalent in patients with HF, reported in up to 49% of HF population.19 In patients with CKD, before and after initiation of haemodialysis, correction of ID with high doses of intravenous iron leads to a significant clinical benefit, resulting in a reduction with 44% of HF hospitalisations.20

WCHF is characterised by a dynamic evolution,1 with ‘stable’ and decompensated periods, mostly secondary to congestion, and existing evidence highlights the prognostic impact of residual congestion at discharge.2 21 Although anaemia is correlated with congestion and rehospitalisations,22 little is known about the impact of ID on in-hospital course of congestion or residual congestion at discharge.23 Our findings showed that patients with ID+ hospitalised for WCHF have more clinical congestion and more severe NYHA class at discharge, characteristics that translate into worse postdischarge outcomes (deaths and rehospitalisations).

NPs as markers of increasing filling pressure24 have distinct trajectories according to ID status. Patients with ID+ were discharged with higher absolute value and less in-hospital decrease of NT-proBNP, despite of prolonged duration of intravenous diuretic therapies. The higher proportion of renal dysfunction seen in patients with ID+ may explain the higher diuretic usage, but may be also an important factor involved in pathophysiology of ID.19 A more intense diuretic regimen, in terms of higher doses or prolonged duration, may be associated with more adverse reactions and clinicians should explore the cause of the insufficient diuretic response rather than extending indiscriminately intravenous diuretic therapy.25

The predischarge phase of hospitalisation represents an opportunity to improve long-term prognosis26, and the results of AFFIRM-AHF trial provide evidence to support the correction of ID predischarge or early after recent HF hospitalisations.27 This is probably even more appropriate in Romania, where patients are faced with limited access to ambulatory settings.28 ID should be corrected promptly and clinicians should not wait for the ambulatory phase to correct ID, in order to avoid exposure to negative effects of in-hospital congestion and to prevent clinical events occurring early after discharge. A delayed approach for ID treatment should be limited only for specific conditions (cancer, infections and systemic inflammation). This approach is supported by the current study since the prevalence of ID did not differ between discharge and 30-days follow-up visit.

In FERIC-RO, the ID+ group had a higher number of events (all-cause mortality and HF rehospitalisations) at 30 and 90 days, a finding distinct from other studies enrolling patients with chronic HF with any LVEF but lower NT-proBNP thresholds for inclusion.29 These findings underline the prognostic significance of hospitalisation for HF and might suggest that ID is not only a biological marker of severity in AHF but also a contributor to HF progression.

Our study has several strengths. FERIC-RO was the first Romanian study that addressed ID in patients hospitalised for WCHF and offered a comprehensive description of its prevalence and clinical characteristics. We identified a multiparametric correlation between the presence of ID and congestion and found that patients with ID+ had insufficient decongestion and had worse prognosis in the early months after discharge.

Limitations

This study has also some limitations. A small sample size may not be representative in a larger context, and therefore, its findings might not be generalised with other populations. It is unknown if this cohort represents an accurate view of the Romanian population with worsening HF, as differences exist between academic hospitals and smaller, local hospitals; patients appeared younger, but with more cardiovascular and non-cardiovascular comorbidities compared with data from contemporary registries. As one of the inclusion criteria was being alive at the moment of discharge, in-hospital mortality during the index hospitalisation was not evaluated in the current study. Proportions of patients lost to follow-up at 30-day and 90-day visits limit the analysis of the events and insight into their clinical trajectory. Due to the non-interventional nature of the study, no intravenous iron therapies were administered during hospitalisation or the follow-up period; however, this approach was highly encouraged for each investigator after the 90-day visit. Despite several clinically relevant associations, any inference about causality remains speculative due to the observational nature of the study.

Conclusions

In patients hospitalised for WCHF, ID is a frequent comorbidity. ID was associated with persistent clinical and biological congestion and with high event rates in the first 3 months after discharge. Since the prevalence and laboratory characteristics of ID did not differ significantly between discharge and 30-day follow-up, a strategy targeting early correction of ID may reduce postdischarge event rates.