Myocardial Iron Deficiency and Mitochondrial Dysfunction in Advanced Heart Failure in Humans

Abstract

Background

Myocardial iron deficiency (MID) in heart failure (HF) remains largely unexplored. We aim to establish defining criterion for MID, evaluate its pathophysiological role, and evaluate the applicability of monitoring it non‐invasively in human explanted hearts.

Methods and Results

Biventricular tissue iron levels were measured in both failing (n=138) and non‐failing control (NFC, n=46) explanted human hearts. Clinical phenotyping was complemented with comprehensive assessment of myocardial remodeling and mitochondrial functional profiles, including metabolic and oxidative stress. Myocardial iron status was further investigated by cardiac magnetic resonance imaging. Myocardial iron content in the left ventricle was lower in HF versus NFC (121.4 [88.1–150.3] versus 137.4 [109.2–165.9] μg/g dry weight), which was absent in the right ventricle. With a priori cutoff of 86.1 μg/g d.w. in left ventricle, we identified 23% of HF patients with MID (HF‐MID) associated with higher NYHA class and worsened left ventricle function. Respiratory chain and Krebs cycle enzymatic activities were suppressed and strongly correlated with depleted iron stores in HF‐MID hearts. Defenses against oxidative stress were severely impaired in association with worsened adverse remodeling in iron‐deficient hearts. Mechanistically, iron uptake pathways were impeded in HF‐MID including decreased translocation to the sarcolemma, while transmembrane fraction of ferroportin positively correlated with MID. Cardiac magnetic resonance with T2* effectively captured myocardial iron levels in failing hearts.

Conclusions

MID is highly prevalent in advanced human HF and exacerbates pathological remodeling in HF driven primarily by dysfunctional mitochondria and increased oxidative stress in the left ventricle. Cardiac magnetic resonance demonstrates clinical potential to non‐invasively monitor MID.

Nonstandard Abbreviations and Acronyms

CAT

catalase

COX I/II/III/IV

Complexes I/II/III/IV

CS

citrate synthase

DCM

dilated cardiomyopathy

DHE

dihydroethidium

DMT‐1

divalent metal transporter 1

ETC

electron transport chain

FPN

ferroportin

GPX

glutathione peroxidase

GR

glutathione reductase

GSH

reduced glutathione

GSSG

oxidized glutathione

HELP

Human Explanted Heart Program

HF‐MID

heart failure with myocardial iron deficiency

HF‐NID

heart failure without myocardial iron deficiency

HOPE

Human Organ Procurement and Exchange Program

LVIDd/LVIDs

LV internal dimensions at end‐diastole/end‐systole

MDA

Malondialdehyde

MID

myocardial iron deficiency

NFC

Non‐failing control

NYHA

New York Heart Association Classification

ROS

reactive oxygen species

SID

systemic iron deficiency

SOD

superoxide dismutase

TEM

transmission electron microscopy

TFR‐1

transferrin receptor 1

Clinical Perspective

What Is New?

• Myocardial iron deficiency is common in explanted failing human hearts with either dilated cardiomyopathy or coronary artery disease.

• Myocardial iron deficiency correlated with greater adverse myocardial remodeling, oxidative stress, and suboptimal mitochondrial structure and function.

• Myocardial iron deficiency correlated with reduced levels of iron importers, transferrin receptor‐1 and divalent metal transporter‐1, and increased levels of the sole iron exporter, ferroportin, in the sarcolemma.

What Are the Clinical Implications?

• Magnetic resonance imaging can detect myocardial iron deficiency.

• Iron supplementation in patients with myocardial iron deficiency, in the absence of major systemic iron deficiency and anemia, is a potential therapy for patients with advanced heart failure.

Heart failure (HF) remains extremely prevalent on a global scale with high morbidity and mortality. ¹ , ² Comorbid conditions in patients with advanced HF not only complicate the presentation and treatment, but also play an instrumental role in progression of HF. ³ Thus, management of comorbidities is gaining equal importance to treating the primary cause of HF itself. ² , ⁴ Iron deficiency (ID) is the most common malnutrition globally, and often co‐exists with HF irrespective of the presence of abnormal blood cell indices (i.e., anemia). ⁵ , ⁶ Conventionally, systemic iron deficiency (SID), defined as either absolute deficit with serum ferritin <100 μg/L or functional insufficiency combining serum ferritin 100–300 μg/L and transferrin saturation <20%, represents ID in the context of HF. ³ , ⁶ , ⁷ , ⁸ Prior studies have demonstrated the detrimental impact of SID on patients’ physical capacity and clinical outcomes, ⁵ , ⁹ whereas iron supplementations exhibited substantial benefits constituting a promising therapeutic target. ² , ⁴ , ¹⁰ However, the diagnosis of SID is solely relied on circulating hematopoietic markers, and screening for SID in patients with HF without anemia remains uncommon. Emerging evidence has highlighted the presence of myocardial iron deficiency (MID) in several HF cohorts, and consistently revealed a weak association with systemic iron status. ¹¹ , ¹² , ¹³

The heart has the highest metabolic demand and is fueled largely by mitochondrial activity. ¹⁴ , ¹⁵ , ¹⁶ Iron is an important micronutrient whose role extends beyond oxygen transport and erythropoiesis to cellular energetics in mitochondria and oxidative stress homeostasis. ¹⁷ Mechanistically, intracellular iron availability is maintained by iron trafficking pathways including uptake, utilization, storage, and excretion of the iron by cardiomyocytes. ¹⁸ While systemic iron regulation is a critical determinant of erythropoiesis and anemia, homeostatic iron levels in the heart are controlled at the tissue level. ¹⁹ , ²⁰ However, the direct burden of MID in patients with HF is unknown. As such, elucidating the prevalence and mechanism of MID, and its impact on mitochondrial function and anti‐oxidative protection directly from the human failing myocardium is clearly warranted.

Accordingly, we studied the burden of MID in the largest cohort of explanted human hearts to date and determined its pathophysiological implications on the failing hearts. Furthermore, we explored the suitability of using quantitative parametric mapping with cardiac magnetic resonance (CMR) as a non‐invasive imaging modality to assess myocardial iron levels. Our results revealed a high prevalence of MID in diseased human explanted hearts which correlated with worsened clinical status and adverse remodeling. We showed greater mitochondrial damage and loss of function in the setting of MID, which was associated with overall reduced expressions of major iron importers. Magnetic resonance imaging provided a useful tool to assess myocardial iron levels, possibly guiding a precision medicine‐based approach to iron supplementation therapy. Taken together, our data revealed that MID is highly prevalent in advanced HF and worsens mitochondrial function, and thereby identifying an unappreciated role for correcting MID in patients with HF.

Methods

The data that support the findings of this study are available from the corresponding author upon reasonable request. Refer to Data S1 for expanded methods.

Human Explanted Hearts: Tissue Procurement and Preparation

Heart specimens from the non‐failing control (NFC, n=46) and adult failing hearts (HF, n=138) were procured from the Human Organ Procurement and Exchange (HOPE) program and Human Explanted Heart Program (HELP), respectively. Our diseased cohort consisted of patients with end‐stage HF secondary to coronary artery disease (CAD, n=67) or dilated cardiomyopathy (DCM, n=71) who underwent heart transplantation. The NFC hearts were obtained from brain dead donors with no past history of major comorbidities or cardiovascular diseases, and antemortem echocardiography demonstrated normal ejection fraction of the left and right ventricles as well as normal LV dimensions. ²¹ , ²² , ²³ , ²⁴ , ²⁵ Informed consent was obtained from all patients and both programs conformed to the ethical principles of the Declaration of Helsinki, and were approved by the institutional review committee and Health Research Ethics Board at the University of Alberta, Edmonton, Canada. Clinical data were obtained by chart review.

Heart tissue procurement strictly followed our well‐established protocols. ²¹ , ²² , ²³ , ²⁴ , ²⁵ Transmural myocardial samples from both ventricles were obtained by avoiding the epicardial fat and scar tissues. For this study, mid‐anterior ventricular walls from both LV (approximately two‐thirds below the aortic valves) and right ventricle (RV, approximately two‐thirds below the tricuspid valves) were procured from the NFC and DCM failing hearts, whereas peri‐infarcted and non‐infarcted regions from LV were collected from failing hearts with CAD involving the left anterior descending artery (LAD). All the full‐thickness specimens were snap‐frozen and/or OCT‐mounted frozen in liquid nitrogen, and then stored in the −80 ℃ freezers for subsequent molecular and histochemical analyses. ²⁴

Tissue Iron Level Measurement

Chamber‐ and pathogenesis‐specific myocardial tissue iron levels were directly measured by inductively‐coupled plasma resonance mass spectrometry as previously described at the Department of Pathology and Laboratory Medicine, London Health Sciences Center, and St. Joseph’s Health Care, London, Western Ontario. ²⁶ , ²⁷ , ²⁸ Measurement of myocardial iron content was carried out from both ventricles in non‐ischemic DCM and NFC hearts, while the levels from peri‐ and non‐infarction regions in LV were anatomized in relation to LAD blockade. Tissue samples were analyzed in triplicate and the average values were reported in this study.

Spectrophotometric Assays for ETC Enzymes

Supernatant from the left ventricle (LV) homogenate was used to assess the electron transport chain (ETC) enzyme activity of NADH:ubiquinone oxidoreductase (COX I), succinate dehydrogenase (SDH, COX II), decylubiquinol cytochrome c oxidoreductase (COX III), NADH cytochrome c oxidoreductase (COX I+III), succinate cytochrome c reductase (COX II+III), cytochrome c oxidase (COX IV), and citrate synthase (CS). ²⁹ Enzyme activity (nmol·min⁻¹·mg⁻¹) was normalized to volume and protein concentration, following protein determination with Bradford assay. Furthermore, the reaction specificity was assured by subtracting the inhibitor‐resistant activity from the total enzymatic activity, which were conducted in parallel. The inhibitor for COX I (1mM rotenone), COX II (1 M malonate), COX III (1 mg/mL antimycin A), COX I+III (1 mmol/L rotenone), COX II+III (1 M malonate), and COX IV (10mM KCN) were added to each corresponding reaction mixture prepared separately. ²⁹ Measurements were performed in triplicate.

Spectrophotometric Assays for Antioxidant Enzymes

Sample homogenates from flash‐frozen LV tissues were prepared as previous described and all measurements were repeated in duplicate and the average value was used. ²⁶ , ³⁰

Catalase Enzyme Assay

Catalase (CAT) activity was measured according to the method described previously with minor modification. ³¹ , ³² Specific activity (units/mg) was defined as the rate of H₂O₂ consumption per minute per milligram of protein sample.

Superoxide Dismutase Enzyme Assay

Superoxide dismutase (SOD) activity was assayed based on the competition for O₂ ⁻ between (ferri‐)cytochrome c and SOD following its spontaneous dismutation with minor modifications. ³¹ , ³³ One unit of activity was defined as the amount of SOD required to inhibit the initial reduction rate of ferri‐cytochrome c by 50%. Mitochondrial SOD (SOD2, Mn/Fe‐SOD) activity was determined by adding 100 mmol/L KCN to a matched reaction mixture prepared from the same sample. The overall Cu/Zn‐SOD activities from cytosol (SOD1) and extracellular matrix (SOD3) were completely inhibited by the KCN (100 mmol/L) added. ³⁴

Glutathione Peroxidase Enzyme Assay

Glutathione peroxidase (GPX) activity was measured based on the oxidation of reduced glutathione (GSH) by GPX coupled to the disappearance of NADPH catalyzed by glutathione reductase (GR). ³¹ , ³⁴ The rate of NADPH oxidation was monitored spectrophotometrically at 340 nm. Briefly, 2 assays (A & B) were prepared each containing 0.1 M K₂HPO4/1 mmol/L EDTA (pH 7.0), 10 mmol/L L‐glutathione reduced (G4251, Sigma, MO), 2.4 unit/mL glutathione reductase (G3664, Sigma, MO). The non‐enzymatic and H₂O₂‐independent NADPH depletion were subtracted from the total GPX activity, by comparing the absorbance changes after addition of H₂O₂ in the assays. Activities were normalized to the added lysate volume and protein concentration.

Measurement of Myocardial Oxidative Stress

Malondialdehyde Assay

Myocardial malondialdehyde (MDA) levels were assayed using a commercially available colorimetric kit in accordance with the manufacturer’s instructions (Abcam, ab233471). The total concentration of free MDA (μmol/L per mg) was determined by reference to the MDA standard curve correcting for the sample lysate dilution as well as total amount of protein loaded. ²⁶ , ³⁵ Each sample was measured in duplicate, with the average value reported.

Glutathione Recycling Assay

Total myocardial glutathione, including the reduced (GSH), and oxidized (GSSG) forms, and their redox ratio (GSH:GSSG) were quantitated by the enzymatic recycling method with minor modification. ²⁷ , ³⁰ , ³⁶ Each sample was analyzed in triplicate, and the average value was finally adopted in our study.

Dihydroethidium Staining and Densitometry

In situ generation of reactive oxygen species (ROS) was determined by incubating the 5–10 µm cryosections with dihydroethidium dye (DHE, D1168, Invitrogen), following the application of TrueBlack Lipofuscin Quencher (#23007, Biotium). The superoxide, as the redox indicator, was fluorescently visualized red within nucleus under Olympus IX81 fluorescence microscope. Quantitative measurements of DHE fluorescence intensity, corrected by the average pixel intensity from the background, were further carried out using MetaMorph software (Basic version, 7.7.0.0, Molecular Devices, Inc.). ²⁶ , ³⁵ , ³⁷ , ³⁸ , ³⁹

Subcellular Fractionation and Western Blot

Subcellular fractionations were carried out as previously described with modifications. ⁴⁰ The purity of each fraction was further validated by using anti‐rabbit TLR‐4 (Santa Cruz, sc‐10741; membrane marker), anti‐rabbit Caspase‐3 (Cell Signaling, 9662S; cytosolic marker), and anti‐rabbit Histone H3 (Cell Signaling, 4499s; nuclear marker). ⁴¹ Western blot was performed on flash snap‐frozen human myocardium tissues as we previously published. ⁴⁰ , ⁴¹ The below primary antibodies were used: anti‐rabbit TFR‐1 (Cell Signaling, 13208s); anti‐rabbit FPN (Novus, NBP1‐21502); anti‐rabbit FTN (Abcam, ab75973); anti‐mouse DMT‐1 (Abcam, ab55735), followed by incubation with HRP‐conjugated secondary antibodies at 1/5000 dilution (Cell Signaling). The total protein loadings were visualized by MemCode reversible stain (24585, Thermo Scientific) as a loading control. Fiji ImageJ software (NIH, Bethesda, MD) was used for band intensity quantitation.

Histological Analysis and Confocal Microscopy

The 5 μm thick sections of the formalin‐fixed paraffin‐embedded tissue were stained with picrosirius red (PSR) and Masson’s trichrome stain for morphometric analyses as described previously. ²⁶ , ²⁷ From each heart, n=2 sections were stained with n=20–25 random images analyzed from each section in a blinded manner. Cardiomyocyte cross‐sectional area was evaluated as using fluorescence wheat‐germ agglutinin staining previously published. ²¹ , ²⁶ , ⁴⁰ From each heart, n=2 sections (including one technical control) were examined, with n=20–25 random images captured from each section in a blinded manner. Within each image, n=25 cardiomyocytes were unbiasedly sampled from whole regions (4 corners and center) into our analyses.

Non‐specific autofluorescences (mainly lipofuscin) from the human OCT‐embedded blocks were eliminated by applying TrueBlack Lipofuscin Quencher (#23007, Biotium) to the cryosections, followed by standardized tissue fixation, deparaffinization, antigen retrieval and permeabilization. The sections were then incubated with primary antibodies overnight as per manufacturer instructions, followed by incubation with fluorochrome conjugated secondary antibodies (Invitrogen). Intracellular protein colocalizations were acquired under laser scanning confocal microscopy (Leica TCS SP5, Leica Microsystems), and quantitative analyses were performed using Fiji ImageJ software. ⁴⁰

Transmission Electron Microscopy

Fresh transmural myocardium from LV (<1 mm³) were promptly fixed in 2% glutaraldehyde upon explantation. The post‐fixative samples were immersed in solution of 2% uranyl acetate (UA) and 0.1 M sodium acetate (pH 5.2) for high‐contrast en bloc staining, followed by dehydration using graded ethanol and acetone solutions, and immediate infiltration with Spurr resin (Leica Electron Microscopy Sciences, Hatfield, PA). Two resin blocks per sample were sectioned along the longitudinal axis of myofilaments to produce 4 non‐consecutive ultrathin sections (70 μm), which were further post‐stained with 4% UA and 4% lead citrate.

Four 100 μm² regions were randomly selected to obtain n=1 image at 2000×, n=4 images at 4000×, and n=6 images at 10000× resolutions per section for a total of 44 images per sample (H7650, Hitachi, Tokyo, Japan). We established a scoring system evaluating the presence and severity of intramitochondrial inclusions, mitochondrial cristae quality as well as sarcomeric integrity, in which a higher score signified a greater severity of dysfunction (Table S1). Blinded assessment of all images was randomly carried out in triplicate by 2 examiners, and a third adjudicator was involved should any discrepancies arise between the grading.

CMR Imaging

Frozen myocardium from the middle of interventricular septum were adopted to evaluate the tissue iron content by CMR mappings. Based on LV iron level, n=10 and n=4 samples were retrospectively included in the NFC group and each HF subgroup, respectively. However, the subsequent sample preparation, image acquisition, and analytical processing were conducted in a double‐blinded manner. Possible interferences from specimen dimension, environment temperature or surrounding buffer heterogeneity were eliminated by strictly following same sample preparation. ⁴²

CMR experiments were performed on a 3T MRI scanner (MAGNETOM Prisma; Siemens Healthcare; Erlangen, Germany) with body coil excitation and a 2.5 cm surface coil for signal reception. Longitudinal relaxation time (T₁) images were acquired with a saturation‐recovery gradient‐echo pulse sequence with the following parameters: 10 slices (no gap), 1 mm slice thickness, 30 mm by 60 mm field of view, 128 phase‐encoding, and 256 readout points for 0.23 mm in‐plane spatial resolution. Saturation‐recovery images with a recovery time of TS=1000 ms and full recovery were used to calculate T₁ in each pixel. Transverse relaxation time (T₂) images were acquired with a spin‐echo sequence with identical spatial coverage and resolution as the T₁ acquisition, with echo‐times of TE=11 ms in steps of 11 ms to 88 ms. T₂* images were acquired with a multi‐echo gradient‐echo sequence with identical spatial coverage and resolution as the T₁ and T₂ acquisitions. Averaged relaxation values (measured in msec) from all pixels within each tissue sample were automatically selected for analyses; all measurements were completed in duplicate.

Statistical Analysis

The normality of data distribution and homogeneity of variance were firstly assessed by Shapiro‐Wilk test and Levene test, respectively. Continuous variables were presented as medians with interquartile ranges (median, Q1–Q3) for clinical parameters, or means±standard deviations (mean±SD) for experimental measurements. Categorical data were summarized as numbers with percentages (integer, %). One‐way ANOVA (followed by Tukey post hoc analysis), or independent sample t‐test was used to compare continuous variables between groups, while Mann‐Whitney U test or Kruskal Wallis test was applied for non‐parametric comparisons as appropriate. All categorical data were analyzed by Chi‐squared test or Fisher’s exact test where applicable. Continuous data sets were visualized by box plots with overlapping data points, or bar charts (upper line of the bar represents mean value) in a consistent manner. Pearson’s correlation or Spearman rank correlation was used to evaluate the statistical association between variables of interest, including parametric and non‐parametric variates, respectively. Multiple linear regression models were performed to estimate the relationship between 2 or more explanatory variables and the dependent variable, including the logistic regression algorithm for binary outcome prediction.

Briefly, multiple linear regression model was performed (Table S2) to explore the estimated coefficients of multiparametric CMR (including T1, T2, and T2*) and HF pathogenesis in predicting myocardial iron content. In this model, the myocardial iron level was the outcome variable, whereas the 3 CMR mapping sequences (measured in msec) and one etiological category (1: non‐failing control; 2: heart failure) were the predictor variables. The “Enter” method (direct entry) was accepted for the variable selection in this linear regression model. Data visualization and graphical representation was performed on Origin for Windows, Version 2018b (OriginLab Corp., M.A.). IBM SPSS Statistics for Windows, Version 21 (IBM Corp., N.Y.) was used for data analysis and narrative interpretation. A two‐tailed P value <0.05 was considered statistically significant.

Results

Prevalence of Myocardial Iron Deficiency and Its Association With Clinical Characteristics in Patients With End‐Stage Heart Failure

We examined the chamber‐specific myocardial iron levels in explanted human hearts which included a total of 46 non‐failing donor hearts and 138 failing hearts with a primary pathogenesis of DCM (n=71) or CAD (n=67) (Table; Table S3). The LV had higher myocardial iron content than the RV in both NFC [LV: 137.4 (109.2–165.9) versus RV: 95.1 (77.6–121.5) μg/g dry weight, P<0.001] and HF [LV: 121.4 (88.1–150.3) compared with RV: 96.40 (73.5–120.0) μg/g d.w., P<0.001] groups (Figure 1A). Surprisingly, iron level decreased only in the LV of patients with HF (P=0.015) with similar changes seen between the 2 etiological cohorts, while no difference was observed in the RV between NFC and HF groups (P=0.648) indicating MID is a major insult to the systemic ventricle (Figure 1A). Accordingly, we defined MID with a priori threshold<86.1 μg/g d.w. in LV, based on its distinct distribution pattern between non‐diseased and failing hearts (Figure 1B). Our tissue‐based approach clearly separated NFC from the HF cohort resulting in n=32 (23%) failing hearts classified as iron‐deficient for the first time (Figure 1B). Our analyses also revealed that MID is LV‐specific and subsequent molecular investigations were all performed using LV samples.

Table 1.

Baseline Clinical Characteristics of Patients With Normal Myocardial Iron Levels Versus Myocardial Iron Deficiency

	HF‐NID (N=106)	HF‐MID (N=32)	P value
Clinical
Age, y	54.5 (47.0–61.75)	54.5 (41.8–60.3)	0.904
Sex, Male	92 (87)	24 (75)	0.110
Pathogenesis, DCM	53 (50)	18 (56)	0.535
BMI, kg/m2	26.3 (24.2–30.2)	27.7 (24.6–30.6)	0.631
Physical assessment
SBP, mm Hg	100.0 (88.0–120.0)	100.0 (90.0–115.0)	0.865
DBP, mm Hg	64.0 (56.0–72.0)	63.0 (51.5–73.0)	0.535
Electrocardiography
QRS duration, ms	122.0 (92.0–151.0)	136.0 (104.8–164.0)	0.180
AF	19 (18)	8 (24)	0.377
LBBB	12 (11)	3 (10)	0.757
Echocardiography
LA volume index, mL/m2	43.5 (32.6–61.9)	47.3 (30.0–57.8)	0.944
LVPWT, mm	8.9 (8.0–10.0)	8.5 (8.2–10.0)	0.952
RVSP, mm Hg	35.8 (26.6–47.9)	34.1 (28.2–37.2)	0.298
TAPSE, mm	1.4 (1.0–1.8)	1.4 (1.1–1.8)	0.952
RVd Basal, cm	4.2 (3.6–4.9)	4.2 (3.9–4.8)	0.484
Blood parameters
Ferritin, μg/L	142.5 (61.3–309.0)	91.0 (47.0–151.0)	0.103
Serum iron, μmol/L	10.0 (8.5–13.0)	10.5 (6.3–14.0)	0.667
TIBC, μmol/L	54.0 (47.0–63.0)	61.0 (53.3–72.0)	0.093
sTF (%)	19.5 (13.8–27.3)	17.0 (9.3–19.8)	0.150
Hemoglobin, g/L	128.5 (111.0–140.0)	125.5 (108.5–135.3)	0.424
MCV, fL	90.0 (86.0–95.0)	89.5 (86.8–92.3)	0.757
MCHC, g/L	336.0 (330.0–342.0)	332.5 (322.5–338.5)	0.067
eGFR, mL/min per1.73 m2	55.0 (41.0–70.0)	58.5 (47.0–76.8)	0.384
Devices
Pacemaker	65 (61)	21 (66)	0.660
ICD/BiV‐ICD	86 (81)	27 (84)	0.676
VAD	67 (63)	20 (63)	0.942
Medications
ACEi/ARB	81 (76)	28 (88)	0.177

ACEi indicates angiotensin converting enzyme inhibitors; AF, atrial fibrillation; ARB, angiotensin receptor blockers; BiV‐ICD, bi‐ventricular implantable cardioverter‐defibrillator; BMI, body mass index; DBP, diastolic blood pressure; DCM, dilated cardiomyopathy; eGFR, estimated glomerular filtration rate based on MDRD equation; LBBB, left bundle‐branch block; LVPWT, left ventricular posterior wall thickness; MCHC, mean corpuscular hemoglobin concentration; MCV, mean corpuscular volume; NYHA, New York Heart Association Functional Classification; RVd Basal, basal right ventricular diameter; RVSP, right ventricular systolic pressure; SBP, systolic blood pressure; sTF, saturation of transferrin; TAPSE, tricuspid annular plane systolic excursion; and TIBC, total iron binding capacity. Categorical variables reported by count with percentage in parenthesis: sex, pathogenesis, diagnosis of AF and LBBB, device implantation, and medications. Continuous variables reported by median with interquartile range in parenthesis: age, BMI, physical assessment, QRS duration, echocardiography, and blood parameters.

Figure 1. Myocardial iron deficiency in failing explanted human hearts with chamber‐specific features of iron levels.

A, Myocardial iron levels in the left (LV) and right ventricles (RV) in non‐failing control (NFC, n=46) and explanted failing hearts (HF, n=138). B, Distribution of myocardial iron levels and the definition of myocardial iron deficiency (MID) with a cutoff value of 86.1 µg/g dry weight separating out 23% of patients with HF. C, NYHA classification, LVEF, LVIDs and LVIDd in HF‐NID (n=106) versus HF‐MID (n=32). HF‐MID indicates patients with HF with myocardial iron deficiency; HF‐NID indicates patients with HF without myocardial iron deficiency; LVEF, LV ejection fraction; LVIDd, LV end‐diastolic internal diameter; LVIDs, LV end‐systolic internal diameter; and NYHA, New York Heart Association. *P<0.05, **P<0.01 compared with NFC; ^# P<0.05, ^## P<0.01 compared with HF‐NID.

Our NFC group consisted of 46 donors (male: 50%), with a median age of 47.0 years (28.0–56.5), heart weight of 350.0 grams (312.0–427.0), and LV ejection fraction (LVEF) of 60% (52.5%–62.5%) in the absence of major comorbidities and cardiovascular diseases. The HF‐MID had comparable demographics, comorbid and cardiovascular history, hemodynamic parameters, and medical therapy as the patients with HF with normal myocardial iron levels (HF‐NID, Table 1). We found no correlations between myocardial iron content and hemoglobin (r=0.017, P=0.84), serum ferritin (r=0.028, P=0.82), and systemic iron levels (r=0.173, P=0.23) in the HF cohorts indicating that in the absence of distinct systemic iron deficiency (SID) and severe anemia, MID is a highly localized to the heart (Figure S1). Importantly, HF‐MID patients had remarkably higher NYHA class (II/III/IV, HF‐MID: 0/15.6/84.4 versus HF‐NID: 7.5/34.9/57.5%, P=0.017), and worsened myocardial remodeling and systolic function as reflected by greater lowering in LVEF (HF‐MID: 17.1 [11.5–29.1] versus HF‐NID: 20.4 [15.0–31.3]%, P=0.04) and larger increase in LV internal dimensions at end‐systole (LVIDs, HF‐MID: 58.0 [47.0–67.0] versus HF‐NID: 50.0 [38.3–60.8] mm, P=0.03) and end‐diastole (LVIDd, HF‐MID: 67.0 [51.5–71.5] versus HF‐NID: 58.0 [50.3–67.0] mm, P=0.03) (Figure 1C).

Myocardial Iron Deficiency Is Linked to Adverse Myocardial Remodeling and Mitochondrial Dysfunction

We next examined the cellular and subcellular characteristics of the explanted failing hearts (LV) to correlate the effects of MID on adverse myocardial remodeling. Increased myocardial hypertrophy and interstitial fibrosis are prominent features of heart diseases. ⁴³ , ⁴⁴ Cardiomyocyte size assessed by WGA staining displayed greater increase in the failed hearts regardless of cardiac iron status compared with NFC (NFC: 239.6±46.4 versus HF: 670.0±248.6 μm², P<0.001), which was consistent with our gravimetric analysis of the explanted hearts (NFC: 363.0±74.3 versus HF: 472.4±130.0 gram, P<0.001) (Figure 2A through 2C; Figure S2A). Meanwhile, PSR staining captured strikingly higher fibrosis in the failing hearts which was further exacerbated in the HF‐MID group (Figure 2D and 2E; Figure S2B), and it was confirmed by classic Masson’s trichrome staining (Figure S2C). Next, we developed a comprehensive scoring system evaluating cardiomyocyte ultrastructure, presence and severity of intramitochondrial inclusions and mitochondrial cristae quality (Table S1). Myofilament disarray and severe lysis were identified in the HF‐MID group in both DCM and CAD samples, in comparison with NFC (Figure 2F; Figure S2D and S2E). Qualitative assessment of the mitochondria using TEM showed severe distortion of cristae and increased inclusion bodies in the HF‐MID group with occasional mitochondrial lysis (Figure S3). These features demonstrated exacerbation of the adverse remodeling in the explanted failing human hearts with MID.

Figure 2. Adverse myocardial remodeling in failing hearts is exacerbated by myocardial iron deficiency.

A through C, Representative wheat germ agglutinin staining (scale bar=100 µm) (A), cardiomyocyte cross‐sectional area across diseased subgroups (n=20 for NFC, n=10 for each HF subgroup) (B), and heart weight (n=46 for NFC; n=106 for HF‐NID and n=32 HF‐MID groups) (C) showing greater hypertrophy in the explanted failed hearts compared with non‐failing controls (NFC). D and E, Representative picrosirius red staining (D) and relative quantification (E) of myocardial fibrillar content exhibiting significantly higher fibrosis in the iron‐deficient failing hearts. n=20 for NFC, n=10 for each HF subgroup. F, Representative TEM images illustrating myofilament disarray, derangement and lysis in the HF subgroups identified from both DCM and CAD samples (scale bar=500 nm; asterisks represent areas with severe alterations). CAD‐NI indicates non‐infarcted from coronary artery disease; CAD‐PI, peri‐infarcted from coronary artery disease; DCM, dilated cardiomyopathy; MID, myocardial iron deficiency; NFC, non‐failing controls; NID, no myocardial iron deficiency; and TEM, transmission electron microscopy. *P<0.05, **P<0.01 compared with NFC; ^# P<0.05, ^## P<0.01 compared with HF‐NID.

A central role of iron relates to the superoxide dismutase (SOD) family which provides a key defense against superoxide radicals (O₂ ⁻) produced in the mitochondria as a byproduct of the respiratory chain activity (Figure 3A). We assessed total SOD and mitochondria‐specific (SOD2) enzymatic activities. While the enzymatic activities of total (NFC: 127.4±22.1 versus HF: 131.8±71.2 unit/mg protein per min, P=0.786) and mitochondrial (NFC: 69.9±16.8 versus HF: 72.5±50.8 unit/mg protein per min, P=0.822) isoforms were comparable between the 2 groups, we further identified a marked suppression of their ROS‐scavenging capacities in the HF‐MID group (Figure 3B), with concordant changes seen across all etiological subgroups (Figure 3C and 3D). In contrast, functional activities of 2 downstream antioxidant enzymes, catalase (CAT) and glutathione peroxidase‐1 (GPX1), were not affected by myocardial iron levels (Figure S4A and S4B). Overall, myocardial reduced glutathione (GSH) levels and GSH to oxidized glutathione (GSSG) ratio were further decreased in iron‐deficient failing hearts (Figure 3E and 3F) likely due to the loss of SOD‐related antioxidant protections. ²⁶ , ³⁴ While markedly elevated oxidative stress within failing myocardium was delineated by DHE staining, the corresponding densitometric analyses confirmed higher level of superoxide associated with myocardial iron insufficiency across all subgroups (Figure 3G through 3I). Additionally, we observed aggravated lipid peroxidation profiles specifically in iron‐deficient failing hearts, consistent with increased oxidative damage in these hearts (Figure 3J and 3K).

Figure 3. Greater oxidative stress in iron deficient failing explanted hearts.

A, Schematic of the mitochondrial function and role of iron in enzymes involved in oxidative stress. B through D Superoxide dismutase (SOD) activity assays showing overall reduced myocardial antioxidant capacity based on total (B, C) and mitochondrial (B, D) SOD activities. n=20 for NFC; n=12 for each HF subgroup with n=36 for HF‐MID and HF‐NID, respectively. E and F, Reduced glutathione (GSH, E) and reduced/oxidized glutathione (GSH/GSSG, F) ratio in HF samples further exacerbated in the MID group. n=10 NFC; n=18 each for HF‐NID and HF‐MID groups. G through I Representative dihydroethidium‐stained images (G) and corresponding densitometries (H, I) delineating the markedly elevated oxidative stress (reflected as the total superoxide levels) in the failing human hearts, which was further inflamed by myocardial iron insufficiency. n=12 for NFC; n=6 for each HF subgroup with n=18 for HF‐MID and HF‐NID, respectively. Scale bar=100 μm. J and K Quantitative colorimetry of the total free malondialdehyde levels highlighting the aggravated lipid peroxidation in iron‐deficient failing myocardium (J), which could consistently be alleviated by restoring iron levels in patients with HF with DCM or CAD (K). n=10 for NFC; n=5 for each HF subgroup with n=15 for either HF‐MID or HF‐NID.

CAD‐NI indicates non‐infarcted from coronary artery disease; CAD‐PI, peri‐infarcted from coronary artery disease; DCM, dilated cardiomyopathy; MID, myocardial iron deficiency; and NID, no myocardial iron deficiency. *P<0.05, **P<0.01 compared with NFC; ^# P<0.05, ^## P<0.01 compared with HF‐NID.

Iron is a component of iron‐sulfur (Fe‐S) clusters in the electron transport chain (ETC) pathways, which in combination with increased oxidative stress and ultrastructural abnormalities, implicates dysfunctional respiratory and metabolic activities in the mitochondria (Figure 4A). Complex‐specific activity assays showed a selective loss of complex I, II, IV (Figure 4B and 4C), and citrate synthase (CS, Figure 4D and 4E) activities in HF‐MID with no decline in complex III functional activity (Figure S4C). Moreover, the enzyme activities of ETC complex I (r=0.54, P<0.001), II (r=0.41, P=0.012), and IV (r=0.47, P=0.004) and CS (r=0.59, P<0.001) within the citric acid cycle correlated positively with myocardial iron levels in failing hearts, providing further evidence for a direct relationship between myocardial iron status and mitochondrial enzyme activities (Figure 4F). These results demonstrated that lowered myocardial iron in human hearts was closely associated with unfavorable myocardium remodeling, and could further exacerbate oxidative stress and deplete respiratory chain activity in the setting of advanced HF.

Figure 4. Assessment of electron transport chain pathway and the impact of myocardial iron deficiency.

A. Schematic of the electron transport chain (ETC) for ATP production and its role in the generation of reactive oxygen species (ROS). B and C, Reduction in enzymatic activities of complex I, II, III, and IV in HF samples (top panel) with greater functional decrease seen in complex I, II and IV in HF‐MID (bottom panel) (B), which were further stratified based on pathogenesis (C). A greater reduction in complex I and II activities were observed in DCM and CAD samples with MID, while reduced complex IV activity was restricted to iron‐deficient samples with CAD. n=10 for NFC, n=36 for HF; n=6 for each HF subgroup with n=18 for HF‐MID and HF‐NID, respectively. D and E, Decreased citrate synthase (CS) activity in HF samples (left panel) compared to NFC, and it was exacerbated in HF samples with MID (right panel) (D), which was primarily driven by lowered CS activity in the iron‐depleted peri‐infarct and non‐infarct regions from CAD (E). n=10 for NFC, n=36 for HF; n=6 for each HF subgroup with n=18 for HF‐MID and HF‐NID, respectively. F, Linear regression analyses showing the strong dependence of complex I, II, IV, and CS enzyme activities on myocardial iron levels within HF cohort. n=36 for each enzyme, and dotted lines represent 95% CI. CAD‐NI indicates non‐infarcted from coronary artery disease; CAD‐PI, peri‐infarcted from coronary artery disease; COX I‐IV, complexes I‐IV; DCM, dilated cardiomyopathy; MID, myocardial iron deficiency; and NID, no myocardial iron deficiency. *P<0.05, **P<0.01 compared with NFC; ^# P<0.05, ^## P<0.01 compared with HF‐NID.

Role of Iron Trafficking System in Myocardial Iron Deficiency

Myocardial iron homeostasis is orchestrated by a tightly controlled regulatory system involving transcriptional control of the iron regulatory protein (IRP‐1/‐2) axis and corresponding changes in key iron transporters, namely transferrin receptor‐1 (TFR‐1), divalent metal transportort‐1 (DMT‐1), and ferroportin (FPN) (Figure 5A). ¹⁷ , ¹⁹ We showed that total levels of TFR‐1 and DMT‐1 were reduced in HF‐MID compared with iron‐sufficient failing hearts by 54.3% and 31.4%, respectively, while the overall level of FPN remained unchanged between 2 groups (Figure 5B; Figure S5). Next, we investigated the intracellular translocation of iron transporters between cytosolic and plasma membrane compartments respectively (Figure 5C through 5K), following our validated subcellular fractionation protocol (Figure S6A and Figure S6B). Likewise, our immunoblotting analysis of major iron importers demonstrated that HF‐MID had significantly reduced membrane and cytosolic levels of both TFR‐1 (Figure S6C, Figure S7) and DMT‐1 (Figure S6D, Figure S8) when compared with their iron‐sufficient counterparts. Interestingly, there was relatively increased expression of FPN on sarcolemma without noticeable change seen in the cytosolic fraction (Figure S6E, Figure S9) of iron‐deficient failing myocardium. The subcellular distribution and shift of iron transporters was further examined across all HF subgroups (Figure 5C through 5K) and similar findings of TFR‐1 (Figure 5C and 5D; Figure S7), DMT‐1 (Figure 5F and 5G; Figure S8), and FPN (Figure 5I and 5J; Figure S9) were observed in DCM and CAD samples with MID, indicating that restricted iron uptake with concomitant iron efflux can underlie the basis of MID in failing hearts. We next used immunofluorescent staining of the iron transporters in DCM and CAD samples to confirm the subcellular location and concordance of changes seen in the Western blot analysis (Figure 5E, 5H, 5K; Figure S10). Collectively, our results demonstrated a differential subcellular regulation of iron transporters in MID characterized by reduced membrane and cytosolic levels of iron uptake transporters, TFR‐1 and DMT1, and increased levels of the iron exporter, FPN, in the membrane.

Figure 5. Assessment of iron transporters using immunoblotting analysis and immunofluorescence staining with confocal microscopy.

A, Schematic of the key iron transporters and the role of the iron regulatory proteins in a cardiomyocyte. B, Immunoblotting analysis (left panel) and quantification (right panel) of the total transferrin receptor 1 (TFR‐1), divalent metal transporter 1 (DMT‐1), and ferroportin (FPN) levels showing overall decreased TFR‐1 and DMT‐1 expressions in HF‐MID, after normalized to NFCs. n=3 for each HF subgroup with n=9 for HF‐MID and HF‐NID, respectively. C through E, Subcellular immunoblotting analysis with representative bands (C) and quantification (D) of TFR‐1 showing overall reduced expressions in both membrane and cytosolic fractions in samples with MID, which was confirmed by immunofluorescence staining (E). F through H, Similarly, subcellular immunoblotting analysis with representative bands (F) and quantification (G) of DMT‐1 showing overall reduced expressions in both membrane and cytosolic fractions in HF subgroups with MID, which was supported by immunofluorescence staining (H). I through K, Subcellular western blot analysis (I) and quantification (J) of FPN, and immunofluorescence staining (K) showing its relative increase in membrane fraction but not in cytosol in samples with MID. CAD‐NI indicates non‐infarcted from coronary artery disease; CAD‐PI, peri‐infarcted from coronary artery disease; Cyto, cytosol; DAPI, 4′,6‐diamidino‐2‐phenylindole; DCM, dilated cardiomyopathy; Mem, membrane; MID, myocardial iron deficiency; NID, no myocardial iron deficiency; and WGA, wheat germ agglutinin. Arrowheads indicated membrane colocalization while asterisks refer to the cytosolic location of the proteins. n=10 for NFC; n=6 each for HF subgroups for all subcellular immunoblotting analyses. *P<0.05, **P<0.01 compared with NFC; ^# P<0.05, ^## P<0.01 compared with HF‐NID.

Myocardial Iron Levels Assessed by CMR Imaging

In the absence of overt SID, we evaluated quantitative imaging using CMR relaxation time constants as a non‐invasive assessment of myocardial iron status. In our study, we explored the feasibility of applying T1, T2, and T2* evaluations in tissue samples using CMR technique (Figure 6A). Our analyses showed that significantly elevated T2 (Figure 6B) and T2* values (Figure 6C) were featured in iron‐deficient failing hearts, with similar changes seen between healthy control and diseased groups. However, T1 signal did not exhibit such distinct alteration (Figure 6B). Multivariate analysis incorporating T1, T2, T2*, and pathogenesis showed moderate predictability of myocardial iron levels providing a reliable and non‐invasive methodology to determine myocardial iron levels (Figure 6D; Table S2). These results demonstrated the ability of CMR to detect and accurately reflect myocardial iron levels as a promising clinical surrogate in patients with advanced HF.

Figure 6. Magnetic resonance imaging of explanted human heart samples in relation to myocardial iron levels.

A through C, Typical parametric maps (T₁, T₂ and T₂*) from a representative sample are illustrated (A) allowing for the quantitative assessment using T1 and T2 (B) and T2* values (C) in non‐failing controls (NFC) and HF subgroups consist of DCM and CAD. D, Multiple linear regression analysis using T1, T2, T2* and pathogenesis as covariates showing a moderately strong relationship with myocardial iron levels (r=0.581; P=0.015; n=34). NID=no iron deficiency; MID=myocardial iron deficiency. n=10 for NFC, n=24 for HF; n=4 for each HF subgroup with n=12 for HF‐MID and HF‐NID, respectively. *P<0.05, **P<0.01 compared with NFC; ^# P<0.05, ^## P<0.01 compared with HF‐NID.

Discussion

Elucidating and treating comorbidities in patients with HF remains a pivotal approach to minimize morbidity and mortality. While systemic iron deficiency and anemia in acute and patients with chronic HF are associated with worsened clinical outcome, iron supplementation in these patients improved prognosis. ³ , ⁷ , ⁴⁵ , ⁴⁶ , ⁴⁷ Current guidelines for the diagnosis and treatment of HF clearly endorse a class I recommendation of iron deficiency and anemia screening in all patients. ² , ⁴ Given that the dominant mechanism of iron regulation occur at the tissue level, myocardial iron homeostasis can be uncoupled from systemic iron profile. ⁸ , ⁴⁵ , ⁴⁸ However, the primary determinant of the myocardial iron levels in failing human hearts remains unexplored. Our prospective human explanted heart program has provided a valuable resource with extensive collection of explanted human heart specimens and clinical phenotypes, thereby facilitating the examination of myocardial iron regulation in relation to advanced HF. ²⁴ , ²⁵ We performed the largest translational study examining myocardial iron levels directly from the LV and RV of explanted human hearts in patients with DCM or CAD.

Our results established that MID in advanced failing hearts is associated with greater adverse remodeling including interstitial fibrosis and cardiac hypertrophy, as well as worsened LVEF and NYHA classification. Interestingly, myocardial iron levels in RV were coherently lower compared with LV regardless of pathogenesis, and the lack of MID in the RV indicated divergent adverse remodelling progression in those patients with HF. ⁴⁹ We also showed that LV‐specific MID was associated with further suppressed ROS‐scavenging capacity, excessive oxidative stress, and impaired mitochondrial respiratory function and altered ultrastructure integrity. While our results were correlative in nature, they were consistent with prior studies directly linking iron deficiency to impaired human cardiomyocyte function and mitochondrial respiration. ¹⁵ , ⁵⁰ Human embryonic stem cell‐derived cardiomyocytes depleted of iron affected mitochondrial function through reduced activity of the Fe‐S cluster‐containing complexes I, II and III with reduced ATP levels and contractile force. ¹⁵ Fe‐S clusters are ubiquitous cofactors composed of iron and inorganic sulfur, which are required for the proper function of Fe‐S proteins involved in a wide range of biological activities, including electron transport in respiratory chain, micronutrient (i.e., iron) sensing, DNA repair and a key component of antioxidant enzymes.

The steady‐state level of myocardial iron is maintained by the concerted action of major iron importers, TRF‐1 and DMT‐1, and the sole iron exporter, FPN. ¹⁷ , ²⁰ , ⁵¹ , ⁵² , ⁵³ The expressions of TRF‐1 and DMT‐1 are positively regulated by the nuclear transcriptional factors, IRP‐1 and IRP‐2, and cardiac selective disruption in the IRP‐1/‐2 axis led to MID and cardiac dysfunction. ¹⁹ , ²⁰ Surprisingly, there was reduced membrane translocation of TRF‐1 and DMT‐1 likely a result of the defective IRP‐1/‐2 pathway in HF‐MID, which clearly was inadequate to restore myocardial iron status. Disruption in IRP‐1/‐2 action would conversely lead to increased translation of FPN mRNA. Indeed, we observed relatively increased membrane fractions of FPN facilitating iron efflux from the failing cardiomyocytes, which is supported by findings in a genetic murine model. ⁵² Suppression of FPN in the membrane compartment provides a promising foundation for correcting MID in patients with advanced HF.

The exact mechanisms by which patients with HF develop MID are not completely understood. An interplay between increased sympathetic activation and iron deficiency was recently observed in patients with chronic HF, suggesting the latter could be more than a comorbidity but a critical component leading to HF. ⁵⁴ Our data supported the hypothesis that iron deficiency is an integral pathophysiology promoting the progression to advanced HF, and that low iron storage in patients with HF is independently associated with escalated rates of mortality and re‐hospitalizations. ⁵⁵ In addition, our results were consistent with clinical trial findings showing that the treatment of iron deficiency irrespective of anemia was beneficial. ⁴⁷ Regulation of skeletal muscle energetics also represents an important mechanism by which iron supply confers benefits in HF in addition to its central role in myocardial iron homeostasis. ⁵⁶

Different therapeutic possibilities embrace iron replacement by oral or IV routes. Several clinical trials with IV iron in chronic patients with HF with SID, have demonstrated equally efficacious and similar favorable safety profiles following correction of iron levels, irrespective of anemia. ¹⁰ , ⁵⁶ Thus, iron status should be assessed in symptomatic patients with HF both with or without anemia, and treatment of iron deficiency warrants consideration in clinical practice. ⁴⁷ The use of CMR to diagnose myocardial iron‐overload is a valid and established technique. ⁵⁷ Our ability to extend the evaluation of myocardial iron levels using CMR to iron‐deficient hearts provides a unique tool to potentially diagnose and monitor patients with HF with MID. Our findings illustrate the potential for precision medicine and correction of MID, especially given the widespread clinical availability of parametric mapping with CMR and its assessment of myocardial iron levels. ⁴² , ⁵⁷ , ⁵⁸ , ⁵⁹ Animal models with selective MID and cardiac dysfunction are corrected by the IV administration of iron supporting the distinct possibility of this approach in patients. ²⁰ In patients with heart failure and iron deficiency, IV ferric carboxymaltose administration changed T2* and T1 cardiac MRI parameters indicative of myocardial iron repletion, further supporting the utility of cardiac MRI to monitor myocardial iron deficiency and its response to therapy. ⁵⁹

There are a few limitations of our study that need to be acknowledged. First, our investigation was cross‐sectional and all patients included were at end‐stage HF, and thus we captured a single time point of the entire disease course thereby limiting the ascertainment of causative relationship when interpreting the experimental results. Secondly, while our non‐failing control hearts demonstrated no evidence of adverse remodeling characteristic of heart failure, they do not represent the true normal myocardium in vivo for reasons including, but not limited to, antemortem medications and metabolic alteration in relation to initial injuries, and postmortem adrenergic storm associated with brain death. In order to minimize these limitations, we used a large number of samples coupled with a comprehensive profile of clinical parameters including comorbidities, past medical history, and treatments, which were integrated into specimen assessment, subgrouping, and data interpretation.

Sources of Funding

This work was supported by the Canadian Institutes of Health Research [CIHR, PJ 44352] and Heart & Stroke Foundation [177654] to GYO. This work was supported, in part, by CIHR project grant [FRN 156393] to JMS. HZ is supported by the China Scholarship Council (CSC) Award.

Disclosures

None.

Supporting information

Data S1

Tables S1–S3

Figures S1–S10