Effects of Intravenous Iron Replacement Therapy on Cardiovascular Outcomes in Patients with Heart Failure: A Systematic Review and Meta-Analysis

(1) Background: Iron deficiency (ID) is an important adverse prognostic marker in patients with heart failure (HF); however, it is unclear whether intravenous iron replacement reduces cardiovascular mortality in this patient group. Here, we estimate the effect of intravenous iron replacement therapy on hard clinical outcomes following the publication of IRONMAN, the largest trial in this field. (2) Methods: In this systematic review and meta-analysis, prospectively registered with PROSPERO and reported according to PRISMA guidelines, we searched PubMed and Embase for randomized controlled trials investigating intravenous iron replacement in patients with HF and co-existing ID. The primary outcome was cardiovascular mortality and secondary outcomes were all-cause mortality, hospitalizations for HF and a combination of the primary outcome and hospitalizations for HF. (3) Results: A total of 1671 items were identified and after removal of duplicates we screened titles and abstracts of 1202 records. Some 31 studies were identified for full-text review and 12 studies were included in the final review. The odds ratio (OR) for cardiovascular death using a random effects model was 0.85 (95% CI 0.69 to 1.04) and for all-cause mortality it was 0.83 (95% CI 0.59 to 1.15). There was a significant reduction in hospitalizations for HF (OR 0.49, 95% CI 0.35 to 0.69) and the combination of hospitalizations for HF and cardiovascular death (OR 0.65, 95% CI 0.5 to 0.85). (4) Conclusions: This review supports the use of IV iron replacement reducing hospitalization rates for HF, however more research is required to determine the effect on cardiovascular mortality and to identify the patient population most likely to benefit.


Introduction
Iron is an essential micronutrient required for fundamental cellular processes including regulation of proteins, lipids and ribonucleic acids supporting oxygen transport and storage, mitochondrial function, muscle metabolism as well as immune and neural function [1][2][3][4][5]. The most frequently used definition of iron deficiency (ID) in patients with cardiovascular disease is ferritin < 100 µg/L or ferritin 100-299 µg/L and transferrin saturation (TSAT) < 20% [6][7][8], however recent studies suggested that relying on ferritin for diagnosis of iron deficiency may be suboptimal [9,10].
ID is highly prevalent in patients with cardiovascular diseases [11] and in particular heart failure (HF) and is an important predictor of adverse clinical outcomes [6]. Whilst oral iron supplements are available, these are poorly absorbed, leading to incomplete iron repletion, and have often significant gastrointestinal side effects, thereby limiting their usefulness. This is particularly important in the heart failure population as iron absorption may be further decreased and compliance reduced due to pill burden [12]. However, intravenous iron replacement has been shown to improve prognosis including quality of life, exercise capacity and hospitalizations for HF [6,7,[13][14][15]. As a consequence, international guidelines for HF suggest screening patients for ID, with or without anemia, and to consider intravenous iron replacement in patients who are found to be deficient [16,17]. However, the recommendations differ in their strength. Whilst the European guidelines recommend that iron replacement should be considered to improve symptoms, exercise capacity and quality of life (QOL) and may be considered to improve hospitalizations, the AHA/ACC/HFSA guidelines only recommend that it is reasonable to use intravenous iron replacement to improve functional status and QOL. Thus, solidifying the available evidence on the effectiveness of intravenous iron replacement may lead to stronger recommendations for iron replacement in the heart failure population.
The underlying mechanisms of how ID affects cardiovascular function are still being elucidated. Recent studies have elegantly demonstrated a potential role for mitochondrial dysfunction to help explain adverse outcomes in patients with HF and co-existing ID. Through use of magnetic resonance spectroscopy (MRS) it has been shown that energetic stores in both skeletal as well as heart muscle are reduced in patients with ID [18,19]. This is important as the effects of iron replacement on quality of life and exercise capacity may be caused by improvement in either improved skeletal function, heart muscle function, or both. However, the beneficial effects of iron replacement on cardiovascular mortality are not well established. In this systematic review and meta-analysis, we assessed the effects of intravenous iron replacement on cardiovascular mortality, all-cause mortality as well as hospitalizations in patients with HF following completion of IRONMAN, the largest randomized controlled trials (RCT) in this field [15].

Materials and Methods
The systematic review and meta-analysis were conducted according to the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) guidelines [20] and was prospectively registered with PROSPERO (CRD42023384715).

Type of Studies, Outcomes and Inclusion and Exclusion Criteria
We planned to only include RCTs comparing intravenous iron replacement to placebo or no treatment, in patients with heart failure and co-existing ID. We included studies reported in full-text or abstract form if sufficient data were available. We included all participants with a diagnosis of HF irrespective of etiology who were also diagnosed as iron deficient by any classification.
The primary endpoint was cardiovascular mortality, secondary endpoints were allcause mortality, hospitalizations for HF, and the combined endpoint of cardiovascular mortality and admission for HF. Studies had to report on either cardiovascular mortality or all-cause mortality to be included in this review. We did not examine studies that did not include a placebo or no treatment arm.

Search Methods
We searched PubMed and Embase from inception until December 2022 without language restrictions or filters. The following search term was used: ((ferric derisomaltose OR ferric carboxymaltose OR iron sucrose OR iron dextran) AND heart failure AND Random*). Reference lists of key studies included in the systematic review were screened for additional references.

Data Collection and Analysis
After removal of duplicates, three reviewers (J.R., N.C. and V.B.) independently screened titles and abstracts. Full text articles were assessed for inclusion in the metaanalysis based on the presence of cardiovascular or all-cause mortality and based on the inclusion and exclusion criteria above by two independent reviewers (J.R. and V.B.). Any disagreements were resolved through discussion or if this was not possible by a third review author (V.S.V.). Covidence was used to document the selection process and to allow completion of a PRISMA flow diagram. A PRISMA checklist (Supplementary Materials) has been completed (Supplemental Material).

Data Extraction and Management
Two authors (J.R. and N.C.) extracted the following data from the included studies using a standardized data collection sheet: First author and study name, year published, countries enrolling, number of patients, blinding details, main inclusion criteria, baseline characteristics of patients, definition of ID used, iron preparation used for treatment, follow up duration as well as primary and secondary endpoints.

Risk of Bias Assessment
Risk of bias in all included studies was assessed using Covidence, through ranking of seven domains as 'low risk', 'high risk' or 'intermediate' by two independent reviewers (NC and GM). We planned to assess publication bias visually and by funnel plot if at least ten studies reported on any outcome measure.

Data Synthesis and Quantification
For all outcomes, we collected data and reported odds ratios (ORs) with 95% confidence intervals. Meta-analysis was performed when at least two studies were identified that reported on the same outcome measure. Heterogeneity was assessed both qualitatively through comparing characteristics of the included studies and through the I2 statistic in each analysis with statistical significance defined as p < 0.05. We planned to analyze data using fixed effects model if heterogeneity was low or otherwise using random effects models in Review Manager 5.3.

Subgroup Analysis and Sensitivity Analysis
If possible, we planned to carry out the following pre-determined subgroup analyses: HFpEF vs. HFrEF, anemic vs. non-anemic patients, male vs. female and according to NYHA classification, HF etiology and Age. If enough data were available, we planned to perform a sensitivity analysis excluding studies with two or more domains of high risk of bias. Additionally, we planned to perform a 'leave-one-out' meta-analysis.

Characteristics of Included Studies and Risk of Bias Assessment
Characteristics of all included studies are summarized in Table 1 and the corresponding risk of bias assessment in Figure 2. The studies were heterogeneous in terms of study size and methodological quality. Importantly, the event rate of cardiovascular mortality or all-cause mortality was low, particularly in smaller pilot studies.

Primary Endpoint-Cardiovascular Mortality
Seven studies including 3292 patients reported on cardiovascular mortality or allcause mortality where cardiovascular mortality could be adjudicated from the information provided in the full text articles [6][7][8]15,21,22,26]. Using a fixed effects model, 213 events/1725 patients in the intravenous iron replacement group and 241 events/1567 patients in the control group yielded an odds ratio of 0.85 (95% CI 0.69 to 1.04) ( Figure 3A). No individual study reached statistical significance but, numerically, all studies included here reported odds ratios < 1. Funnel plot analysis was not performed as less than ten studies reported on the primary outcome.

Characteristics of Included Studies and Risk of Bias Assessment
Characteristics of all included studies are summarized in Table 1 and corresponding risk of bias assessment in Figure 2. The studies were heterogeneous terms of study size and methodological quality. Importantly, the event rate cardiovascular mortality or all-cause mortality was low, particularly in smaller p studies.

Primary Endpoint-Cardiovascular Mortality
Seven studies including 3292 patients reported on cardiovascular mortality or all-cause mortality where cardiovascular mortality could be adjudicated from the  all-cause mortality where cardiovascular mortality could be adjudicated from th information provided in the full text articles [6][7][8]15,21,22,26]. Using a fixed effects model 213 events/1725 patients in the intravenous iron replacement group and 241 events/156 patients in the control group yielded an odds ratio of 0.85 (95% CI 0.69 to 1.04) (Figur 3A). No individual study reached statistical significance but, numerically, all studie included here reported odds ratios < 1. Funnel plot analysis was not performed as les than ten studies reported on the primary outcome.

All-Cause Mortality
A total of 11 studies incorporating 2379 patients reported on all-cause mortality [5 8,15,21,23-26,28]. Using a fixed effects model, our meta-analysis estimates an odd ratio o 0.88 (95% CI 0.70 to 1.10) based on 218 events/1284 patients in the interventional arm and 233 events/1095 patients in the control arm ( Figure 3B). Similarly to cardiovascula mortality, event rates for all-cause mortality were modest or small, except for th IRONMAN study [15], which included 1869 patients with a median follow up of 2. years. A funnel plot was not suggestive of significant publication bias ( Figure 3C).

Heart Failure Hospitalizations
We included ten studies (3423 patients) reporting on hospitalizations for hear failure [6][7][8]15,21,22,[24][25][26][27]. Our random effects model estimates an odds ratio of 0.4 (95% CI 0.35 to 0.69) favoring intravenous iron replacement ( Figure 3D). A funnel plo was not suggestive of publication bias ( Figure 3E). Our findings were robust in sensitivity analysis including only studies with one or less high-risk domains identified during risk of bias assessment (data not shown). Similarly, this result was not altered in a leave-one-out sensitivity analysis (data not shown).

All-Cause Mortality
A total of 11 studies incorporating 2379 patients reported on all-cause mortality [5][6][7][8]15,21,[23][24][25][26]28]. Using a fixed effects model, our meta-analysis estimates an odd ratio of 0.88 (95% CI 0.70 to 1.10) based on 218 events/1284 patients in the interventional arm and 233 events/1095 patients in the control arm ( Figure 3B). Similarly to cardiovascular mortality, event rates for all-cause mortality were modest or small, except for the IRONMAN study [15], which included 1869 patients with a median follow up of 2.7 years. A funnel plot was not suggestive of significant publication bias ( Figure 3C).

Heart Failure Hospitalizations
We included ten studies (3423 patients) reporting on hospitalizations for heart failure [6][7][8]15,21,22,[24][25][26][27]. Our random effects model estimates an odds ratio of 0.49 (95% CI 0.35 to 0.69) favoring intravenous iron replacement ( Figure 3D). A funnel plot was not suggestive of publication bias ( Figure 3E). Our findings were robust in a sensitivity analysis including only studies with one or less high-risk domains identified during risk of bias assessment. Similarly, this result was not altered in a leave-one-out sensitivity analysis.

Heart Failure Hospitalizations and Cardiovascular Death
The pre-specified endpoint of cardiovascular death and hospitalizations for heart failure was reported only in three studies [6,15,22]. Of note, the three included studies were the largest in the field including 2704 patients. We estimate an odds ratio of 0.65 (95% CI 0.5 to 0.85) using a random effects model ( Figure 3F). Due to the small number of studies included in this analysis we were unable to perform sensitivity analysis of funnel plot analysis.

Endpoints in Chronic Heart Failure and Long Follow Up
Because mortality after an acute admission with heart failure may be higher, we also conducted additional subgroup analyses including only studies with chronic heart failure as opposed to acute heart failure and in a separate analysis only studies with a long follow up of one year or greater. Including chronic heart failure studies only, the overall results were not altered (see Supplemental Material). Similarly, including only studies with a long follow up did not alter the main results of this study (see Supplemental Material).

Discussion
To our knowledge, this systematic review and meta-analysis is the first to report on outcomes of intravenous iron replacement therapy in patients with HF and co-existing ID since the completion of the IRONMAN trial [15]. In the present analysis, we focused on hard clinical outcomes and in particular our primary outcome cardiovascular mortality but also all-cause mortality, hospitalizations for heart failure and a combination of hospitalizations and cardiovascular mortality. Our findings suggest a lack of sufficient evidence to conclude whether intravenous iron replacement decreases cardiovascular mortality in patients with HF and ID. All studies included in this systematic review and meta-analysis numerically reported lower cardiovascular event rates in the intravenous iron replacement groups compared to placebo or standard of care; however, the differences are not statistically significant in individual studies or the present meta-analysis. Similarly, there was not enough evidence to conclude on any effect on all-cause mortality with intravenous iron replacement.
Hospitalization for heart failure and the combined endpoint of hospitalizations for heart failure and cardiovascular death were significantly reduced in the intravenous iron replacement group. These secondary endpoints are clinically important for patients, especially their quality of life, and importantly for costs incurred in health care systems. Whilst our study did not demonstrate improved cardiovascular or all-cause mortality with intravenous iron replacement, the benefits we see with hospitalizations have important implications. Overall, our findings suggest that international heart failure guidelines [16,17] could recommend intravenous iron replacement not only to improve functional outcomes but also to reduce hospitalizations.
The underlying mechanisms of how ID and, in turn, IV iron supplementation affect cardiovascular outcomes remain incompletely understood. On a cellular level, ID induces reduced mitochondrial electron transfer chain activity, reduced mitochondrial respiration, a switch towards immature glycolytic metabolism, reduced cellular ATP availability and reduced contractility of human cardiomyocytes [29,30]. Importantly, at least some of the changes occurring during ID are reversible upon replenishment of intracellular iron. Similar effects appear to occur in vivo in human hearts and skeletal muscle as assessed by magnetic resonance spectroscopy [18,19]. These studies have elegantly demonstrated that ID is associated with reduced cardiac energetic stores such as the phosphocreatine (PCr) to ATP ratio. IV iron replacement, in turn, improves phosphocreatine recovery half-time, a measure of energetic stores, at least in skeletal muscle. The effects of IV iron replacement on exercise capacity in patients with heart failure and ID may, therefore, be mediated in part through effects on skeletal and cardiac muscle.
The current study is not without limitations. We used aggregate data from published reports of RCTs to conduct this meta-analysis. Many of the included studies were small in size and the results of this systematic review and meta-analysis were mainly driven by the findings of the largest four trials [6,7,15,22]. Individual patient data meta-analysis is sometimes considered the 'gold standard' of systematic reviews and it would be desirable to perform such an analysis on these four trials; however, this is time-consuming and requires collaboration between various research teams [31].
The included studies also differed in the HF population under study including whether patients had acute or chronic heart failure. For example, by far the largest studies included in this analysis are the AFFIRM-AHF [22] and the IRONMAN [15] studies. The former study recruited patients during an admission for acute heart failure whilst the latter recruited patients with stable chronic HF. Nevertheless, at least for HF hospitalizations, both of these studies support the use of intravenous iron replacement suggesting that this therapy is beneficial across a wide spectrum of HF populations and presentations.
Studies included in this review used various intravenous iron preparations including ferrous carboxymaltose (FCM), ferric derisomaltose (FDM), iron sucrose (IS) and sodium ferric gluconate complex (SFGC) ( Table 1); however, to our knowledge, there are no headto-head comparisons of these preparations available at present. It is therefore unknown, whether any of the preparations are more effective than the others. Most data in the current systematic review and meta-analysis stems from RCTs using FCM or FDM in the treatment arms. Prior to the IRONMAN trial, all larger RCTs had used FCM as the intravenous iron preparation. The additional evidence from the IRONMAN suggests that there may well be a class effect of intravenous iron replacement. However, larger trials of the other preparations (IS and SFGC) would be needed to investigate whether they are equally effective.
We were unable to perform subgroup analysis of different patient groups such as HFpEF and HFrEF, various NYHA severity, HF etiology, patients age, renal function or the presence of absence of anemia as this data could not be extracted in this aggregate meta-analysis. Such an analysis would be desirable in the setting of an individual patient data meta-analysis in the future. For example, using individual patient data, a direct comparison of patients with NYHA severity II and III could be performed, helping to clarify which patients are more likely to benefit. Similarly, comparisons could be made in different HF groups such as HFpEF and HFrEF, according to HF etiology, or other factors such as the time interval from initial diagnosis of heart failure.
Many of the trials included in here relied on the current 'guideline definition' [16] for identification of patients with ID, i.e., ferritin < 100 µg/L or ferritin 100-299 µg/L and transferrin saturation (TSAT) < 20% used in pivotal early trials in the field [6,7]. However, we are becoming increasingly aware that a different definition of ID based on either TSAT measurement or serum iron concentrations rather than ferritin might be more appropriate [9,10]. The rationale for the use of alternative definitions is that ferritin can be released into the circulation during any sort of cell damage and inflammatory processes and indeed have been found to be elevated in patients diagnosed as iron deficient through the gold-standard of bone marrow biopsy [9]. Furthermore, since traditional markers of ID may not always correlate with tissue iron content, cardiac MRI, utilizing specialized techniques like T2*, has been proposed as a technique to assess accurately myocardial iron content [32,33]. Identifying patient populations who are most likely to benefit from intravenous iron replacement could, in theory, improve the efficacy of this therapeutic approach.

Conclusions
This is the largest systematic review and meta-analysis of intravenous iron therapy in patients with HF and co-existing ID to date, and the only one to include the largest trial, IRONMAN [15], in this field. We show that whilst there is a move towards reduced cardiovascular and all-cause mortality in the intravenous iron group, these did not reach statistical significance. However, hospitalization rates for HF as well as the combined endpoint of cardiovascular mortality and HF hospitalizations were significantly reduced in the intravenous iron therapy groups. Our findings are important, as they confirm the appropriate use of intravenous iron in a wide range of patients with HF. However, at the same time, future studies are desirable to identify the phenotype of patients most likely to benefit, as we were unable to make such conclusions based on the present study and the RCTs included in it. Furthermore, it will be important to clarify, in larger studies, whether cardiovascular mortality might also be reduced.