The transcriptional profile of iron deficiency in patients with heart failure: Heme‐sparing and reduced immune processes

Abstract

Aims

Iron deficiency (ID) is highly prevalent in patients with heart failure (HF) and associated with morbidity and poor prognosis, but pathophysiological mechanisms are unknown. We aimed to identify novel biological pathways affected by ID.

Methods and results

We studied 881 patients with HF from the BIOSTAT‐CHF cohort. ID was defined as a transferrin saturation <20%. Transcriptome profiling was performed in whole blood. Identified targets were validated in a human in vitro stem cell‐derived cardiomyocyte ID model utilizing deferoxamine as iron chelator. ID was identified in 554 (62.9%) patients, and 89 differentially expressed genes between ID and non‐ID were identified, of which 60 were up‐ and 29 were downregulated. Upregulated genes were overrepresented in pathways of erythrocyte development and homeostasis. Heme biosynthetic processes were confirmed as relatively upregulated in ID, while iron–sulfur cluster assembly was downregulated. Downregulated processes further included natural killer cell and lymphocyte mediated immunity. In agreement with patient data, cardiomyocyte iron depletion significantly induced the expression of two genes (SIAH2 and CLIC4), which could be normalized upon iron supplementation. Both SIAH2 and CLIC4 are associated with increased mortality in patients with HF (hazard ratio 2.40, 95% confidence interval 1.86–3.11, p < 0.001 hazard ratio 1.78, 95% confidence interval 1.53–2.07, p < 0.001, respectively).

Conclusion

Iron deficiency is associated with the preservation of heme‐related processes at the cost of iron–sulfur clusters. Immune processes are downregulated, uncovering another high energy demand system affected. SIAH2 and CLIC4 might be modifiable factors in the relation between ID and impaired prognosis.

Introduction

Iron deficiency (ID) is prevalent among the majority of individuals diagnosed with heart failure (HF). ¹ , ² Its presence is associated with unfavourable prognosis and more signs and symptoms of HF. ³ , ⁴ Several factors contribute to ID, such as systemic inflammation, diminished dietary intake, and blood loss, but data explaining the high prevalence are lacking. ⁵ , ⁶ Notably, women with severe HF and kidney disease are particularly susceptible to developing ID. In order to mitigate symptoms and reduce HF‐related hospitalizations, patients with ID can be treated with intravenous iron. ⁶ , ⁷ , ⁸ , ⁹ , ¹⁰ , ¹¹ However, the mechanisms by which ID exerts its detrimental effects in individuals with HF remain unknown.

Iron exhibits the ability to transition between ferrous and ferric states, enabling electron acceptance and donation. This property is vital for its involvement in crucial processes such as mitochondrial oxidative phosphorylation, DNA synthesis, and repair. ¹² Within larger protein structures, iron is incorporated into heme or iron–sulfur cluster complexes. Heme‐bound iron is particularly important for systemic (haemoglobin) and cellular (myoglobin) oxygen transport. ¹² Interestingly, the effects of ID are not contingent upon circulating haemoglobin levels. In our previous studies using a haemoglobin independent system devoid of red blood cells, we demonstrated that iron‐deficient human cardiomyocytes in culture experience mitochondrial dysfunction due to reduced iron–sulfur‐dependent mitochondrial complex activity. ¹³ Heme‐dependent mitochondrial enzyme activity remained unaltered. ¹³ In another in vitro study with rat myoblast, mitochondrial dysfunction was also observed when ID was caused by neurohormonal stimuli. ¹⁴ These in vitro findings require validation in patients with HF, additionally it is crucial to investigate the impact of ID beyond cardiac and skeletal myocytes in HF. ¹⁵ , ¹⁶

The utilization of transcriptomics has emerged as an unbiased method to identify novel disease mechanisms and biomarkers. In a recent study by Nath et al., ¹⁷ whole blood expression levels of two key iron regulatory genes, HEPC and TFRC, were found to be associated with cardiovascular mortality in HF patients. In another small study in HF patients, TFRC was found differentially regulated in whole blood of patients with ID. ¹⁸ Building upon these findings in the BIOSTAT‐CHF (A systems BIOlogy Study to TAilored Treatment in Chronic Heart Failure) cohort, our objective was to uncover the pathways affected by ID by analysing the whole blood transcriptome profile from HF patients and modelling ID in human cardiomyocytes.

Methods

Study population

We included HF patients from the BIOSTAT‐CHF cohort. This cohort has been described in full detail elsewhere. ¹⁹ , ²⁰ , ²¹ , ²² In short, the BIOSTAT‐CHF study included patients either hospitalized for HF or presenting with HF in the outpatient setting from 69 hospital centres in 11 European countries predominantly during 2010–2014. Patients were eligible to participate with a left ventricular ejection fraction (LVEF) of ≤40% or, alternatively, B‐type natriuretic peptide or N‐terminal pro‐B‐type natriuretic peptide (NT‐proBNP) levels of >400 ng/L or >2000 ng/L, respectively. Additionally, patients had to receive suboptimal evidence‐based HF treatment (i.e. ≤50% of target dose of angiotensin‐converting enzyme inhibitors, angiotensin receptor blockers and/or beta‐blockers). After study inclusion, treating physicians were encouraged to up‐titrate these drugs during a 3‐month treatment optimization phase. The BIOSTAT‐CHF study was conducted in accordance with the Declaration of Helsinki. Institutional review board approval was obtained in all countries. All patients provided written informed consent prior to any study‐related activities.

Whole blood transcriptomics study design, patient selection and matching

The transcriptomics dataset consisted of whole blood expression data generated using the Affymetrix Human Transcriptome Array (HTA) 2.0. The primary outcome was cardiovascular mortality. Patient selection was performed based on availability, sufficiency and stability of RNA samples as assessed using both Nanodrop and Bioanalyzer (Agilent 2100 Bioanalyzer System). Subsequently, patients were matched in order to provide two groups that were comparable in terms of age (within 2.5 years) and sex (exact match). Of all 2516 patients enrolled in the BIOSTAT‐CHF cohort, whole blood transcriptome profiling was performed in 944 (37.5%) patients. RNA processing and normalization details are described elsewhere. ¹⁷ Of these 944 patients, 881 could be classified as ID or non‐ID based on iron data availability for this study.

Laboratory measurements

Iron parameters were assessed from venous blood. Blood samples were centrifuged at 2500 g for 15 min (4°C) and stored at −80°C afterwards. Samples were never thawed before laboratory analyses. The following blood markers reflecting iron metabolism were assessed on a Roche modular Cobas 8000 using standard methods: serum iron, ferritin, and transferrin. Transferrin saturation (TSAT) was calculated as follows: (72.17 * iron [mg/dl])/transferrin (mg/dl). ²³

Definitions

Anaemia was defined as a haemoglobin level <12 g/dl in women and <13 g/dl in men as per World Health Organization standards. ²⁴ ID was defined as a TSAT <20%. ²⁵ This definition has been validated against the gold standard test for ID (bone marrow iron staining) in HF patients and has been used previously. ⁶ , ²⁶ HF was categorized as either a reduced ejection fraction (HFrEF; LVEF <40%), mid‐range ejection fraction (HFmrEF; LVEF 40–<50%) or preserved ejection fraction (HFpEF; LVEF ≥50%) according to the 2016 European Society of Cardiology guidelines. ²⁷

In vitro cardiomyocyte culture

Detailed methods on the in vitro human cardiomyocyte culture have been published before and are available in the online supplementary material. ¹³ In short, HUES9 human embryonic stem cells (Harvard Stem Cell Institute) were differentiated into cardiomyocytes, which were subsequently enriched using glucose starvation, resulting in >99% pure spontaneously beating cardiomyocytes. To deplete the cardiomyocytes from iron, 30 𝜇M deferoxamine (D9533, Sigma‐Aldrich) was added to the culture medium and refreshed daily for 4 days. For the purpose of restoring iron levels, 5 𝜇g/ml transferrin partially saturated with iron (T8158, Sigma‐Aldrich) was supplemented to the medium for 3 days.

Quantitative polymerase chain reaction

To analyse gene expression, total RNA was isolated using TRI reagent according to the provided protocol (T9424, Sigma). RNA concentrations have been determined with a Nanodrop 2000 (Thermo Scientific), and cDNA was synthesized using the QuantiTect Reverse Transcription kit (205 313, Qiagen). Gene expression analysis was performed by quantitative real‐time polymerase chain reaction using IQ SYBR Green (170–8885, BioRad). The samples were normalized to the reference gene RPLP0. The primers used can be found in online supplementary Table  S1 .

Statistical analysis

Statistical analyses were done in R‐Studio (v. 4.2.0) or Graphpad Prism 8.4.2. ²⁸ In group comparisons, categorical variables were depicted as numbers with percentages. Normally distributed variables were depicted as means ± standard deviation, non‐normally distributed variables as median with the first and third quartile (Q1–Q3). The means for continuous variables were compared by one‐way analysis of variance (ANOVA) or the Kruskal–Wallis test, while categorical variables were compared by the Chi‐squared test. Kaplan–Meier survival curves were compared using the log‐rank statistic. Cox regression models were used to adjust for the effect of covariates and to calculate hazard ratios (HR). Cox proportional hazards assumptions were assessed (using the R‐based Survival and Survminer packages) by visual inspection of Schoenfeld residuals against time plots. To assess independency, multivariable models were adjusted for baseline characteristics previously identified as predictive for all‐cause mortality in BIOSTAT‐CHF, with the addition of TSAT to correct for iron status. ²¹ Variables included were: age, beta‐blocker use, haemoglobin, NT‐proBNP and blood urea nitrogen levels. Baseline tables and univariate analyses were done using the R‐based compareGroups package. ²⁹ In general, a two‐tailed p‐value of <0.05 was considered statistically significant.

Differential expression analysis

Differential expression analysis was performed using logistic regression, using univariable comparisons between patients with and without ID. The model was subsequently adjusted for the predictors of ID (age, sex, diabetes and haemoglobin), as well as for all‐cause mortality. The multivariable model was used for further analyses. All analyses were adjusted for multiple testing based on the Benjamini–Hochberg method. Differentially expressed genes were determined using a cut‐off point of ≥0.4 or ≤−0.4 for log2‐fold change between groups, as well as an adjusted p‐value ≤0.05.

Pathway overrepresentation analyses

Differentially expressed genes were visualized using volcano plots. Overrepresentation analysis of differentially expressed genes as defined previously was performed using the clusterProfiler package for R based on the gene ontology biological process classification. Non‐redundant biological processes were filtered using the simplify function of clusterProfiler and the results were visualized in network form. For selected biological processes, gene set enrichment analyses were performed using the clusterProfiler package and visualized using barcode plots. In both cases, visualizations were created using the enrichplot package.

Gene set enrichment analyses

Subsequent threshold‐free gene set analyses (GSEA) were performed for selected biological processes and plotted using the barcode function in Limma. ³⁰ These included the ‘heme’ and ‘iron–sulfur’ related processes. Genes were identified searching for gene ontology terms containing either ‘heme’ or ‘iron–sulfur’ in their name or description. These gene sets included a total of 175 genes, of which 151 are uniquely part of heme‐related pathways and 22 are uniquely part of iron–sulfur cluster pathways (online supplementary Table  S2 ).

Results

Baseline characteristics

Baseline characteristics, stratified by iron status, are depicted in Table  1 . A total of 881 patients with HF from the BIOSTAT‐CHF cohort were studied. Mean age was 71.2 (± 10.8) and 664 patients were male (75.4%). A total of 625 patients (70.1%) were classified as having HFrEF, and 57 (6.5%) as HFpEF, the remaining 103 patients (11.7%) belonging to the HFmrEF group. The majority of patients were in New York Heart Association (NYHA) class II or III (37.2% and 46.5%, respectively). The median TSAT was 16.8% (10.8–23.9) and ferritin 102 μg/L (47–193).

Table 1.

Baseline characteristics

	Iron‐deficient (n = 554)	Not iron‐deficient (n = 327)	p‐value
Demographics and medical history
Male sex	391 (70.6)	273 (83.5)	<0.001
Age (years)	73.0 [64.0–79.0]	72.0 [64.0–80.0]	0.664
Previous HF hospitalization(s) in last year	167 (30.1)	86 (26.3)	0.254
Hypertension	362 (65.3)	205 (62.7)	0.471
Diabetes	189 (34.1)	86 (26.3)	0.019
LVEF (%)	30.0 [25.0–38.0]	30.0 [25.0–37.0]	0.800
HFrEF	380 (77.9)	245 (82.5)	0.142
HFmrEF	67 (13.7)	36 (12.1)	0.590
HFpEF	41 (8.40)	16 (5.39)	0.151
Anaemia (%)	249 (47.5)	76 (26.9)	<0.001
Signs and symptoms of HF
NYHA class			<0.001
I	8 (1.50)	11 (3.48)
II	173 (32.4)	155 (49.1)
III	279 (52.2)	131 (41.5)
IV	74 (13.9)	19 (6.01)
Orthopnoea present	214 (38.7)	78 (23.9)	<0.001
Pulmonary oedema with rales			<0.001
No	203 (37.4)	196 (62.2)
Single base	85 (15.7)	38 (12.1)
Bi‐basilar	255 (47.0)	81 (25.7)
Extent of peripheral oedema			<0.001
Not present	154 (33.1)	147 (55.7)
Ankle	155 (33.3)	60 (22.7)
Below knee	105 (22.6)	47 (17.8)
Above knee	51 (11.0)	10 (3.79)
Elevated JVP			0.001
No	216 (57.1)	163 (71.2)
Yes	137 (36.2)	60 (26.2)
Uncertain	25 (6.61)	6 (2.62)
Hepatomegaly	77 (13.9)	40 (12.2)	0.548
Third heart tone	51 (9.21)	27 (8.26)	0.722
Cardiovascular medications
Aldosterone antagonist	275 (49.6)	183 (56.0)	0.081
Diuretics	553 (99.8)	326 (99.7)	1.000
Loop diuretics	549 (99.1)	326 (99.7)	0.421
Beta‐blocker	438 (79.1)	289 (88.4)	0.001
ACEi/ARB	387 (69.9)	255 (78.0)	0.011
Digoxin	136 (24.5)	65 (19.9)	0.130
Statin	322 (58.1)	180 (55.0)	0.412
Laboratory data
Haemoglobin (g/dl)	12.7 (1.79)	13.8 (1.80)	<0.001
Haematocrit (%)	38.9 (5.06)	41.3 (5.20)	<0.001
Serum creatinine (μmol/L)	106 [87.0–133]	102 [84.9–124]	0.163
Urea (mmol/L)	11.5 [7.80–18.6]	9.40 [7.00–15.7]	0.003
Sodium (mmol/L)	139 [137–142]	140 [138–142]	0.001
Potassium (mmol/L)	4.20 [3.87–4.60]	4.30 [3.90–4.60]	0.061
Total bilirubin (μmol/L)	14.0 [9.75–22.9]	13.0 [9.10–19.6]	0.062
Glucose (mmol/L)	6.42 [5.65–8.05]	6.00 [5.30–7.20]	<0.001
HbA1c (%)	6.40 [5.90–7.01]	6.00 [5.62–6.70]	0.007
AST (U/L)	25.0 [18.0–34.0]	26.0 [21.0–35.5]	0.052
ALT (U/L)	22.5 [15.0–34.0]	26.0 [18.0–36.5]	0.016
Gamma‐GT (U/L)	50.5 [26.0–108]	52.0 [29.0–96.0]	0.483
Alkaline phosphatase (μg/L)	88.0 [68.0–121]	81.0 [62.0–105]	0.051
HDL (mmol/L)	0.98 [0.80–1.27]	1.17 [0.93–1.40]	<0.001
LDL (mmol/L)	2.30 [1.67–2.90]	2.82 [2.05–3.60]	<0.001
Total cholesterol (mmol/L)	3.90 [3.20–4.70]	4.50 [3.70–5.29]	<0.001
Triglycerides (mmol/L)	1.16 [0.87–1.52]	1.30 [0.95–1.72]	0.014
TSH (mU/L)	1.70 [1.07–2.75]	1.78 [1.20–3.00]	0.400
Calcium (mmol/L)	1.76 [1.45–1.96]	1.79 [1.57–2.05]	0.010
Phosphate (mmol/L)	0.86 [0.68–1.04]	0.82 [0.68–1.01]	0.417
Albumin (g/L)	31.0 [26.0–36.0]	34.0 [29.0–39.0]	<0.001
NT‐proBNP (pg/ml)	3234 [1447–6802]	1811 [833–3996]	<0.001
GDF‐15 (pg/ml)	3266 [2032–5568]	2188 [1490–3708]	<0.001
PCT (pg/ml)	21.3 [7.94–48.8]	11.2 [4.56–26.2]	<0.001
Cystatin C (ng/ml)	14 550 [9984–20 940]	15 788 [11 016–24 424]	0.009
CRP (ng/ml)	17 978 [8537–33 536]	7929 [3583–17 086]	<0.001
Erythrocytes (1012/L)	4.37 [3.98–4.80]	4.45 [4.01–4.85]	0.259
Platelets (109/L)	222 [176–270]	200 [158–246]	<0.001
Leucocytes (109/L)	8.00 [6.70–9.75]	7.40 [6.30–8.76]	<0.001
IL‐6 (pg/ml)	7.10 [4.20–13.5]	3.20 [2.00–5.60]	<0.001
Iron (μg/dl)	33.5 [22.3–44.7]	72.6 [61.5–95.0]	<0.001
Ferritin (μg/L)	79.0 [38.0–164]	133 [75.0–232]	<0.001
Hepcidin (nmol/L)	4.75 [1.30–13.5]	8.05 [4.53–18.1]	<0.001
sTfR (mg/L)	1.68 [1.27–2.31]	1.32 [1.04–1.64]	<0.001
Transferrin (mg/dl)	200 [160–250]	200 [160–240]	0.363
Transferrin saturation (%)	12.3 [8.53–15.9]	26.5 [22.9–33.2]	<0.001

Data are depicted as n (%), median [interquartile range], or mean (standard deviation).

ACEi, angiotensin‐converting enzyme inhibitor; ALT, alanine aminotransferase; ARB, angiotensin receptor blocker; AST, aspartate aminotransferase; CRP, C‐reactive protein; GDF‐15, growth differentiation factor 15; GT, glutamyl transferase; HbA1c, glycated haemoglobin; HDL, high‐density lipoprotein; HF, heart failure; HFmrEF; heart failure with mid‐range ejection fraction; HFpEF, heart failure with preserved ejection fraction; HFrEF, heart failure with reduced ejection fraction; IL‐6, interleukin‐6; JVP, jugular venous pressure; LDL, low‐density lipoprotein; LVEF, left ventricular ejection fraction; NT‐proBNP, N‐terminal pro‐B‐type natriuretic peptide; NYHA, New York Heart Association; PCT, procalcitonin; sTfR, soluble transferrin receptor; TSH, thyroid‐stimulating hormone.

A TSAT <20% was observed in 554 (62.9%) of the patients, which were considered as having ID. Patients with ID were more likely to be female and had more signs and symptoms of disease (including a higher NYHA class, orthopnoea, pulmonary rales and an elevated jugular venous pressure) and a higher NT‐proBNP value (median 3234 vs. 1811 pg/ml, all p < 0.001). Higher glucose and glycated haemoglobin levels, and a higher prevalence of diabetes was observed in patients with ID (all p < 0.05). Patients with ID also had higher levels of biomarkers indicating inflammation: procalcitonin, C‐reactive protein, leucocytes and interleukin‐6 (all p < 0.001). Lower haemoglobin levels (12.7 ± 1.8 vs. 13.8 ± 1.8, p < 0.0001) and a higher prevalence of anaemia (47.5% vs. 26.9%, p < 0.001) were present in patients with ID. No differences were observed in age, LVEF or serum creatinine.

Differentially expressed genes in iron deficiency

Whole blood microarray identified 89 genes that were differentially expressed in patients with ID compared to patients without ID, of which 60 were up‐ and 29 were downregulated (Figure  1 ). The most upregulated gene in ID was delta‐aminolevulinate synthase 2 (ALAS2), followed by tropomodulin 1 (TMOD1), catenin alpha‐like 1 (CTNNAL1), pleckstrin 2 (PLEK2) and spectrin beta, erythrocytic (SPTB) (online supplementary Table  S2 and S3 ). Downregulated genes in HF patients with ID included killer cell lectin‐like receptor C4 (KLRC4), fibroblast growth factor binding protein 2 (FGFBP2), killer cell lectin‐like receptor F1 (KLRF1), perforin 1 (PRF1) and natural killer cell granule protein 7 (NKG7) (Figure  1 and online supplementary Table  S2 and S3 ). We performed linear regression analyses to confirm the correlation of the differentially expressed genes with TSAT as a continuous variable independent of the TSAT 20% cut‐off (online supplementary Table  S4 ). All these genes were confirmed to strongly associate with TSAT levels.

Figure 1.

Volcano plot showing differentially expressed genes between patients with and without iron deficiency. Each dot represents one of the 17 900 genes quantified in the whole blood microarray in 881 patients. On the x‐axis, the log2‐fold change in gene expression is depicted (positive log2‐fold change is higher biomarker expression in patients with iron deficiency; negative log2‐fold change is lower biomarker expression in patients with iron deficiency), while the y‐axis shows the magnitude of the biomarker expression difference as –log10 of the p‐value. Green dots indicate biomarkers that are not significantly different, orange dots represent biomarkers that are significantly different with a log2‐fold change between −0.4 and 0.4, purple dots represent biomarkers that are significantly different and have a log2‐fold change of ≥0.4 or ≤−0.4. Significance was tested using a logistic multivariable model adjusted for multiple testing based on the Benjamini–Hochberg method.

To assess if these findings were independent of other patient characteristics associating with ID, we repeated the analyses and included age, sex, diabetes and haemoglobin levels as covariables (online supplementary Figure  S1 and Table  S5 ) and additionally with all‐cause mortality as covariable (online supplementary Figure  S2 and Table  S5 ). This did not impact the significance level of any of the differentially regulated genes.

Pathway overrepresentation analysis

To identify common pathways and mechanisms shared between the differentially expressed genes, we performed pathway overrepresentation analyses as presented in Figure  2 . We identified two biological processes entitled ‘erythrocyte development’ and ‘erythrocyte homeostasis’ as upregulated.

Figure 2.

Significantly up‐ or downregulated gene expression networks in patients with iron deficiency as compared to no iron deficiency.

The genes that were downregulated in ID relative to non‐ID were overrepresented in immune‐related processes, including ‘(positive regulation of) natural killer cell‐mediated immunity’, ‘lymphocyte‐mediated immunity’, ‘leucocyte‐mediated cytotoxicity’, and ‘cell killing’.

Gene set enrichment analysis

The differentially regulated biological processes encompass major iron utilizing pathways. To gain a better understanding of these major iron utilizing pathways, we performed gene set enrichment analysis on ‘heme’ and ‘iron–sulfur’ pathways, the two main iron utilizers. Figure  3 depicts the barcode plot (data in online supplementary Table  S6 ), with genes downregulated on the left and upregulated on the right. As can be observed, genes involved in heme processes are overly upregulated (adjusted p < 0.01), while genes involved in iron–sulfur cluster processes are overly downregulated (adjusted p < 0.01).

Figure 3.

Gene set enrichment analyses for heme and iron–sulfur associated pathways. Barcode plot, each vertical line depicts a single gene, with the red lines representing genes of the heme pathway (153 genes) and the blue lines representing genes of the iron–sulfur cluster pathway (24 genes). In case a gene is located to the left (blue area) it is downregulated while it is upregulated when it appears on the rights (red area). It can be observed that blue genes (iron–sulfur cluster related) are downregulated and thus cluster mainly to the left, while the opposite is true for the red genes (heme related), which cluster to the right.

Iron deficiency in human cardiomyocytes

We previously determined cardiomyocyte‐specific effects of ID, but novel genes identified in the present study have not yet been studied in human cardiomyocytes. ¹³ To characterize the role of genes not yet associated with ID, we analysed the cellular expression patterns 7 of the strongest upregulated genes which are expressed in human cardiomyocytes (Figure  4 ). Of these seven genes, expression levels of SIAH2 and CLIC4 were significantly increased through ID by 3.8 and 1.8‐fold, respectively (both p < 0.05). These ID‐related effects were reversible as gene expression levels of both genes returned to baseline after 3 days of supplementation with transferrin bound iron (Figure  4 ).

Figure 4.

Gene expression in human cardiomyocytes of upregulated genes in whole blood. Gene expression levels relative to control conditions (ctrl) for iron‐deficient cardiomyocytes (iron deficient) after 4 days of exposure to deferoxamine and iron supplemented cardiomyocytes (transferrin treated) who were first subjected to 4 days of deferoxamine and subsequently treated with transferrin bound iron for 3 days. Biological replicates were 14 ctrl, 12 iron deficient and 8 transferrin treated, 2 technical replicates were performed. *p < 0.05, statistical test is the one‐way ANOVA with a post hoc Tukey's honestly significant difference test. CISD2, CDGSH iron–sulfur domain 2; CLIC2, chloride intracellular channel protein 2; CLIC4, chloride intracellular channel protein 4; CTNNAL1, alpha‐catulin; SIAH2, E3 ubiquitin‐protein ligase siah2; SLC22A4, solute carrier family 22; TMOD1, tropomodulin 1.

Associations with mortality

Iron deficiency is known to be associated with increased mortality, also in this BIOSTAT‐CHF cohort. ⁶ To study the association between the newly identified two genes involved in the pathology of ID, we performed Cox proportional hazards regression analyses on all‐cause mortality. Both SIAH2 (HR 2.40, 95% confidence interval [CI] 1.86–3.11, p < 0.001) and CLIC4 (hazard ratio 1.78, 95% confidence interval 1.53–2.07, p < 0.001) are associated with an increased risk of mortality (online supplementary Table  S7 ). Multivariable analyses using the previously defined BIOSTAT‐CHF risk model were performed to assess independence of major determinants of mortality. We additionally included TSAT in the model to assess independence of iron status itself. SIAH2 remained significantly associated with mortality (HR 1.35, 95% CI 1.11–1.65, p = 0.003), but CLIC4 did not (HR 1.32, 95% CI 0.94–1.85, p = 0.113).

Discussion

In a large set of patients with HF, we observed distinct transcriptional patterns associated with iron status. Specifically, individuals with ID exhibited higher activation of heme pathways while experiencing downregulation of iron–sulfur pathways compared to those without ID. Notably, a wide array of genes expressed in natural killer cells and other leucocytes displayed decreased expression levels in HF patients with ID compared to HF patients without ID (Graphical Abstract). Furthermore, we identified SIAH2 and CLIC4 as systemically upregulated genes in HF patients with ID. These genes showed similar expression patterns in iron‐deficient human stem cell‐derived cardiomyocytes. Upregulation of both genes is associated with increased rates of mortality in patients with HF. Iron supplementation in vitro normalized the transcription levels of these genes in human cardiomyocytes. These results might suggest that iron supplementation could potentially modulate the expression of these genes and improve patient outcomes in HF.

SIAH2 and CLIC4 expression is affected by iron availability and associated with prognosis

We identified the genes SIAH2 and CLIC4 to be induced by ID in HF patients and human cardiomyocytes. Higher expression levels were associated with increased mortality, although this was dependent on other clinical determinants for CLIC4. Suppression of SIAH2 and CLIC4 expression in human cardiomyocytes could be achieved by correction of the iron deficit. SIAH2 is known to mediate specific protein degradation in the cellular response to hypoxia and is thereby involved in regulating the response to hypoxia mainly through apoptosis and transcription and signalling processes. Loss of SIAH2 has been associated with increased levels of the mitochondrial scaffolding protein AKAP121, reduced mitochondrial fission and reduced cell death and infarct size in mice. ³¹

CLIC4 is highly expressed in cardiomyocytes, and especially in endothelial cells of patients with pulmonary hypertension. ³² Chloride channels regulate membrane potential, transepithelial transport and intracellular pH. They are implicated in angiogenesis, cardioprotection from ischaemia–reperfusion and pulmonary hypertension. Downregulation of CLIC4 protected mice from developing pulmonary hypertension in response to hypoxia. ³² In contrast, pharmacological inhibition of the channel has been implicated to result in increased mitochondrial reactive oxygen species. ³² Whether the increased expression of SIAH2 and CLIC4 in ID are protective or detrimental has to be determined in future mechanistic studies.

Transcriptional alterations in iron deficiency versus no iron deficiency

We employed an unbiased statistical approach of the whole blood transcriptome, followed by hypothesis driven in‐depth analyses (Graphical Abstract). Our analyses identified several processes to be altered in HF patients with ID. These potentially affected relevant cellular systems pertained to (1) erythrocytes: relative upregulation of erythrocyte development and heme‐associated processes, (2) cardiomyocytes: relative upregulation of heme processes and downregulation of iron–sulfur cluster processes, and (3) immune cells: relative downregulation of lymphocyte and natural killer cell pathways. These pathways will be discussed in the following paragraphs.

Erythrocyte development and heme pathways

Both erythrocyte development and heme pathways are upregulated in HF patients with ID compared to those without ID. Additionally, ALAS2 was the most significantly upregulated gene in patients with ID. Relative upregulation of ALAS2 and other erythrocyte development genes might indicate a compensatory process aimed to preserve adequate levels of heme, haemoglobin and erythrocytes in ID. This might be a possible explanation for the relative low prevalence of anaemia in HF patients with ID: 35–45% in HF patients with ID, which is only approximately 15% higher compared to those without ID (20–30%). ¹ , ⁶

Additionally, we show a transcriptional increase in heme‐related processes. In our model of cardiomyocyte ID, we previously showed an unaffected function of mitochondrial complexes IV and V, both of which are heme‐dependent, ¹³ which is in line with the preservation of cardiac heme levels in the majority of patients with ID, and in human cardiomyocytes. ¹³ , ³³ Moreover, high levels of heme iron, and ALAS2 gene and protein expression were found in failing hearts. ³³ ALAS2, SIAH2 and SLC40A1 (encoding ferroportin) are regulated by hypoxia and strongly upregulated in HF patients with ID in our study, as are other genes associated with hypoxia such as EPO, EPOR and VEGFA/B/C. Together, this might indicate a hypoxic response in HF patients with ID, potentially responsible for the preservation of heme complexes.

Iron–sulfur synthesis in cardiomyocytes

Our data revealed that the set of genes regulating iron–sulfur cluster synthesis was overly downregulated. Iron–sulfur clusters are the major other iron requiring process next to haemoglobin, essential for many cellular processes including oxidative phosphorylation. In mice with a cardiac specific deletion of iron regulatory proteins I and II, a reduced activity of mitochondrial complex I activity, dependent on iron–sulfur, was observed causing cardiac failure after stress. ³⁴ This could be restored using intravenous ferric carboxymaltose. ³⁴ Similarly, we previously identified reduced complex I–III activity, all iron–sulfur‐dependent, in human cardiomoycytes in vitro when deprived from iron. ¹³ Melenovsky et al. ¹⁵ identified specifically complexes I and III activity to be reduced in patients with HF. They further extended these findings to reductions in aconitase activity, an iron–sulfur containing enzyme of the citric acid cycle. The reduced activity of the citric acid cycle might contribute to the shift from fatty acids to glycolysis, as has been observed in mice with cardiac iron deficiency. ³⁵

Lymphocytes and natural killer cells

A set of genes involved in immune processes, with the focus on lymphocytes and natural killer cells, were found to be relatively downregulated in patients with HF and ID, compared to patients with HF without ID. HF is a chronic inflammatory disease with elevations in the levels of many proinflammatory cytokines measurable in the circulation. Limiting iron availability is an important defense mechanism against invading pathogens, linking high levels of inflammatory markers to ID. ⁶ Also, in this study, patients with ID had higher levels of the inflammatory markers C‐reactive protein, interleukin‐6 and procalcitonin. Despite this apparent relatively high level of inflammation, natural killer cell and lymphocyte processes were downregulated in patients with ID, potentially indicating their need of iron for activation. In the recent AFFIRM‐AHF and IRONMAN trials, iron supplementation resulted in a numerically lower incidence of infections in patients treated with intravenous iron, which is encouraging. ¹⁰ , ¹¹ However, inflammation is complex and regulated by many different cell types of which the individual contribution to HF pathophysiology has not been fully elucidated. Mechanistic studies as well as further intervention studies are needed to shed further light on this paradigm.

Clinical implications

Our findings have several clinical implications. Our data show that iron and haemoglobin levels are not necessarily linked. Patients with ID often have a preserved heme synthesis and relatively normal haemoglobin levels. Therefore, normal haemoglobin levels do not exclude ID or its profound effects on mitochondrial functioning. Patients with HF should thus be screened for ID regardless of their haemoglobin level. ²⁷ Furthermore, our study revealed alterations in immune‐related processes in patients with HF and ID. This suggests that ID may impact immune function in a population already susceptible to chronic inflammatory comorbidities such as obesity and diabetes mellitus. Understanding these immune alterations is particularly important given the fragile nature of HF patients and their increased vulnerability to complications associated with immune dysregulation.

Strengths and limitations

The BIOSTAT‐CHF cohort is a well characterized cohort of patients with HF, and the transcriptomic analyses utilized provided us with an unbiased approach to identify new disease mechanisms. Findings in blood might not directly correlate to changes in tissues, and our findings of erythrocyte and lymphocyte processes illustrate the sensitivity of the analyses for blood present cells. However, iron‐related processes are so fundamental to cellular function that we could subsequently confirm and reverse some of these findings in human cardiomyocytes in vitro. Though a large database, our multivariable survival analyses are performed in 881 patients limiting statistical power. Additionally, the majority of patients had HFrEF, potentially limiting generalizability. Finally, further mechanistic studies are necessary to link the transcriptional changes described to functional alterations.

Conclusion

Our study sheds light on the underlying mechanisms of iron ID in HF patients. We observed that ID is linked to a distinct transcriptional profile characterized by reduced iron–sulfur cluster synthesis and altered immune response, while heme synthesis remains relatively unaffected at the transcriptional level. We identified SIAH2 and CLIC4 as genes that are upregulated in the presence of ID and associated with increased mortality in HF patients. Notably, in human cardiomyocytes, the elevated expression of these genes could be reversed through iron supplementation. These findings provide valuable mechanistic insights into the pathophysiology of ID in HF and underscore the importance of assessing iron status in HF patients, irrespective of their haemoglobin levels. By recognizing and addressing ID, clinicians can potentially improve patient outcomes and optimize HF management strategies.

Funding

BIOSTAT‐CHF was funded by a grant from the European Commission (FP7‐242209‐BIOSTAT‐CHF; EudraCT 2010‐020808‐29).

Conflict of interest: The University Medical Center Groningen, which employs several of the authors, has received research grants and/or fees from Vifor Pharma, AstraZeneca, Abbott, Boehringer Ingelheim, Cardior Pharmaceuticals GmbH, Ionis Pharmaceuticals, Novo Nordisk, and Roche. S.A. has received fees from Abbott, Bayer, Boehringer Ingelheim, Cardiac Dimension, Cordio, Impulse Dynamics, Novartis, Occlutech, Servier, and Vifor Pharma; and has received grant support from Abbott and Vifor Pharma. A.A.V. has received consultancy fees and/or research grants from Amgen, Bayer, Boehringer Ingelheim, Cytokinetics, Merck/MSD, Myokardia, Novartis, Novo Nordisk, and Roche Diagnostics. N.J.S. holds a chair funded by the British Heart Foundation and is a National Institute for Health and Care Research Senior Investigator. All other authors have nothing to disclose.

Supporting information

Appendix S1. Supporting Information.