Iron deficiency in pulmonary vascular disease: pathophysiological and clinical implications

Abstract

Aims

Iron deficiency is common in pulmonary hypertension, but its clinical significance and optimal definition remain unclear.

Methods and results

Phenotypic data for 1028 patients enrolled in the Redefining Pulmonary Hypertension through Pulmonary Vascular Disease Phenomics study were analyzed. Iron deficiency was defined using the conventional heart failure definition and also based upon optimal cut-points associated with impaired peak oxygen consumption (peakVO₂), 6-min walk test distance, and 36-Item Short Form Survey (SF-36) scores. The relationships between iron deficiency and cardiac and pulmonary vascular function and structure and outcomes were assessed. The heart failure definition of iron deficiency endorsed by pulmonary hypertension guidelines did not identify patients with reduced peakVO₂, 6-min walk test, and SF-36 (P > 0.208 for all), but defining iron deficiency as transferrin saturation (TSAT) <21% did. Compared to those with TSAT ≥21%, patients with TSAT <21% demonstrated lower peakVO₂ [absolute difference: −1.89 (−2.73 to −1.04) mL/kg/min], 6-min walk test distance [absolute difference: −34 (−51 to −17) m], and SF-36 physical component score [absolute difference: −2.5 (−1.3 to −3.8)] after adjusting for age, sex, and hemoglobin (all P < 0.001). Patients with a TSAT <21% had more right ventricular remodeling on cardiac magnetic resonance but similar pulmonary vascular resistance on catheterization. Transferrin saturation <21% was also associated with increased mortality risk (hazard ratio 1.63, 95% confidence interval 1.13–2.34; P = 0.009) after adjusting for sex, age, hemoglobin, and N-terminal pro-B-type natriuretic peptide.

Conclusion

The definition of iron deficiency in the 2022 European Society of Cardiology (ESC)/European Respiratory Society (ERS) pulmonary hypertension guidelines does not identify patients with lower exercise capacity or functional status, while a definition of TSAT <21% identifies patients with lower exercise capacity, worse functional status, right heart remodeling, and adverse clinical outcomes.

Structured Graphical Abstract

Structured Graphical Abstract.

Mild PH was defined as mean pulmonary arterial pressure 21–24 mmHg, exercise-induced PH, or mPAP >24 mmHg with pulmonary vascular resistance (PVR) <3 Wood units. WSPH, World Symposium on Pulmonary Hypertension; PH, pulmonary hypertension; mPAP, mean pulmonary artery pressure; TSAT, transferrin saturation; QoL, quality of life; CMR, cardiac magnetic resonance; RHC, right heart catheterization; RV, right ventricle; RAP, right atrial pressure; PA, pulmonary artery; PCWP, pulmonary capillary wedge pressure; PVR, pulmonary vascular resistance; RC, resistance/compliance.

See the editorial comment for this article ‘Redefining both iron deficiency and anaemia in cardiovascular disease’, by J.G.F. Cleland et al., https://doi.org/10.1093/eurheartj/ehad154.

Introduction

Iron is an essential cofactor of iron–sulfur clusters in the electron transport chain of mitochondria, in anti-oxidative enzymes and in tricarboxylic acid cycle enzymes.¹ Iron deficiency is associated with an energetic deficit, especially in high energy–demanding tissue, leading to early skeletal muscle acidification and diminished left and right ventricular (RV) contractile reserve during exercise.^2–7 Right ventricular structural and metabolic remodeling is observed in the setting of iron deficiency.^8,9 Maladaptive RV remodeling forms a hallmark feature associated with increased morbidity and mortality in patients with pulmonary vascular disease (PVD).¹⁰

A number of relatively small observational studies have recognized iron deficiency as a common comorbidity in PVD, with prevalences ranging widely between 3% and 71%. This relates in part to the absence of a uniform definition. The 2022 European Society of Cardiology (ESC)/European Respiratory Society (ERS) guidelines define iron deficiency similarly as in heart failure [ferritin <100 g/mL or 100–299 ng/mL with transferrin saturation (TSAT) <20%],¹¹ although much validation of this definition in the field of PVD has not been performed. The field of heart failure, in which iron deficiency is a causal risk factor for disease progression, has taught the importance of using validated definitions for iron deficiency when selecting patients for iron treatment in randomized controlled trials.^12,13 Like heart failure, some data suggest that iron deficiency, in the setting of PVD, is associated with diminished quality of life (QoL) and exercise capacity.^14–21 Yet confirmation in large studies, optimal definition, and precise delineation of the mechanism (cardiac, pulmonary vascular, and/or peripheral limitation) leading to exercise limitation remain awaited.

The goals of this study were two-fold: (i) to determine and validate the best definition of iron deficiency in PVD and document its prevalence in the entire range of PVD and (ii) to determine the association between iron deficiency and exercise capacity and QoL and determine clinical associations in terms of cardiac and pulmonary vascular remodeling.

Methods

Participant enrollment and sites

Redefining Pulmonary Hypertension through Pulmonary Vascular Disease Phenomics (PVDOMICS) is a prospective multicenter observational study aimed at providing in-depth phenotyping of patients with PVD. Participants were prospectively enrolled between November 2016 and October 2019 at seven centers across the USA [Brigham and Women’s Hospital, Harvard Medical School, Columbia University Irving Medical Center, Weill Cornell Medical Center, Johns Hopkins Hospital, Mayo Clinic (Rochester), University of Arizona (Tucson), and Vanderbilt University Medical Center]. The design, rationale, and baseline features of the PVDOMICS study cohort have been published previously.²² Briefly, patients with pulmonary hypertension (PH) or PVD comparators (see further for definition) were included if (i) >18 years, (ii) clinical indication for right heart catheterization, and (iii) able to perform the entire phenotyping protocol. Both treatment naïve and patients treated with PH-specific drugs were allowed, as such new cases and patients with a history of PVD were included. A list of exclusion criteria is reflected in Supplementary data online, Table S1. Additionally, healthy controls were recruited from the enrolling centers. Data management and core lab adjudication were performed at the Data Coordinating Center at the Cleveland Clinic. The study was supported by the National Institutes of Health/National Heart, Lung, and Blood Institute and registered at ClinicalTrials.gov (identifier NCT02980887). The phenotyping protocol was approved by local institutional review boards, and all participants provided written informed consent.

Patient populations and definitions

Pulmonary hypertension

Participants were classified according to the 2013 World Symposium on Pulmonary Hypertension (WSPH) guidelines, meaning that a mean pulmonary artery pressure (mPAP) ≥25 mmHg was defined as PH.²³  Supplementary data online, Table S2 illustrates the hemodynamic definition and descriptions used to categorize in the five different WSPH categories.

Pulmonary vascular disease comparators

Enrollees with mild elevation in mPAP (21–24 mmHg), exercise-induced PH (exercise mPAP >30 mmHg and ΔmPAP/δcardiac output slope >3 mmHg/L/min), or mPAP >24 mmHg with pulmonary vascular resistance (PVR) <3 Wood units were deemed comparators for patients with overt PH. While the 2022 ESC/ERS PH guidelines also define such patients as PH, enrollment in PVDOMICS predates these guidelines¹¹; nevertheless, these patients were enrolled as a milder phenotype and for the sake of this analysis treated separately as the PVD comparator group. A list of definitions and descriptions of this cohort is reflected in Supplementary data online, Table S3. For both the PH patients and the PVD comparators, we excluded patients who were in WSPH category 5 as they might have hemoglobinopathies and could be mechanistically different when assessing the role of iron deficiency. Finally, when presenting the data for PH and PVD comparators in combination, we refer to the larger group as the PVD spectrum (thus including patients with milder and more advanced PH).

Healthy controls

Included age-, sex-, race-, and Hispanic ethnicity–matched healthy controls aged ≥18 years old.

Data collection

Details of the protocol have been previously published.^22,24 All participants (PH, PVD comparators, and healthy controls) underwent comprehensive clinical phenotyping including collection of demographics, QoL surveys, medication use and comorbidities, 6-min walk testing (6MWT), electrocardiogram, cardiac magnetic resonance (CMR) imaging, routine clinical labs, and cardiopulmonary exercise testing that were leveraged in this analysis. Enrollees with PH or PVD comparators all underwent protocolized right heart catheterization.²⁵ Cores at the Data Coordinating Center performed central reading for electrocardiogram, echocardiogram, CMR imaging, cardiopulmonary exercise testing, and right heart catheterization. Vital status was assessed annually after enrollment with censoring at 22 November 2021.

Iron deficiency and clinical associations

Iron parameters including iron (µmol/L), ferritin (ng/mL), and transferrin (mg/dL) were measured in all participants at the laboratory diagnostic core laboratory of the Cleveland Clinic using the Roche Cobas platform (c501 Module, Roche Diagnostics, Indianapolis, IN). Transferrin saturation, expressed in percent, was calculated from the ratio of serum iron divided by transferrin multiplied by 70.9. Hemoglobin (Hb) levels were measured at the enrolling sites according to institutional standards, and anemia was defined according to the World Health Organization definition of Hb <12 g/dL in women and Hb <13 g/dL in men.

Two definitions of iron deficiency were evaluated. First, the heart failure definition of iron deficiency endorsed by 2022 ESC/ERS PH guidelines being a ferritin between 100 and 299 ng/mL with TSAT <20% or a ferritin <100 ng/mL, irrespective of TSAT was assessed in relation to peak oxygen consumption (peakVO₂) and submaximal exercise (6MWT) capacity and health-related QoL (HRQoL).¹¹ Several studies carried out in patients with heart failure have raised questions in regard to this definition, as patients with ferritin <100 ng/mL, but a TSAT >20% display relatively preserved exercise capacity, functional status, and reduced treatment response with intravenous ferric carboxymaltose.^8,13,26,27 Secondly, we evaluated if a different definition could better identify patients with reduced exercise capacity and functional status. The relation between the definitions of iron deficiency was assessed in patients across the PVD spectrum. The newly validated optimal definition was subsequently used in further exploratory analysis, assessing the impact of iron deficiency on exercise capacity, functional status, and cardiac/pulmonary vascular function.

Statistical analysis

Descriptive results are presented as median with 25th and 75th percentiles or mean ± standard deviation depending on the absence or presence of a normal distribution. Categorical variables were expressed as numbers and percentages. Group comparison was done using Pearson’s χ² or Fisher’s exact test for categorical variables, and t-test or Wilcoxon rank sum test as appropriate. Ordinal variables were compared using the Wilcoxon rank sum test. Analysis for continuous endpoints requiring covariate adjustment was performed using an analysis of covariance (ANCOVA). All models of peakVO₂, 6MWT, and 36-Item Short Form Survey (SF-36) were adjusted for age, sex, and hemoglobin given the biologic relation. All CMR models were adjusted for sex, given sex differences in chamber volumes. The optimal cut-points for the iron parameters (ferritin, iron, and TSAT) in relation to functional status measures (peakVO₂, 6MWT, and SF-36 physical component score) were analyzed by running serial receiver operating characteristic (ROC) curve analyses. Each functional metric value within the 10%–90% range of the distribution was used as a cut-point to bisect the distribution. An ROC analysis was then run for the iron parameters and repeated for each of the functional status measure values. The resulting areas under the curve (AUC) statistics were compared. The iron parameter cutoff corresponding to the highest AUC result became the initial optimal cut-point. Additionally, results were internally validated by repeating this general ROC approach with nested bootstrap samples. One thousand bootstrap samples were generated for each functional status value. Each of these initial samples were further bootstrapped 1000 times, and then optimal cut-points were derived for each of subsample based on the AUC. The corresponding iron parameter cutoffs were averaged. This bootstrap-derived optimal cut-point was then applied to the value sets which were excluded from the original 1000 bootstrap samples. Averages of the resulting ROC statistics represented internally validated results. Subgroup analyses were performed using a generalized linear regression model. Cox proportional hazard models were used to determine the impact of covariates on overall survival. Non-proportionality of hazards was assessed for mortality by adding an interaction of TSAT groups with the logarithm of time to the Cox proportional hazard model. All hypothesis testing was two sided and a P-value <0.05 was set to indicate significance. Statistical analyses were performed using either R or SPSS version 25.

Results

Optimal definition of iron deficiency

Of the 1171 patients enrolled in the PVDOMICS cohort, iron levels were available in 1156 (99%), of whom 90 were healthy controls, 724 were PH patients, and 342 were PVD comparators (Figure 1). Patients with group 5 (as by design) were excluded; therefore, the final study cohort consisted of 1028 patients in the PVD spectrum consisting of 693 PH patients and 335 PVD comparators. Iron deficiency according to the heart failure definition was present in 513 (74%) of patients with PH and 743 (72%) across the PVD spectrum (not reflected in Figure 1). However, this heart failure definition of iron deficiency endorsed by the ESC/ERS PH guidelines did not discriminate between patients with reduced or preserved peakVO₂ [absolute difference: −0.6 (−1.6 to 0.4) mL/min/kg, P = 0.208], 6MWT [absolute difference: −4.3 (−24 to 15) m, P = 0.670], or SF-36 physical component score [−0.55 (−2.0 to 0.9), P = 0.425]. This was related to the fact that patients with iron deficiency based on a reduced ferritin without TSAT reduction displayed relatively preserved functional capacity and QoL (Figure 2). Patients with isolated hypoferritinemia formed 27.7% of the PVD spectrum cohort with iron deficiency, but significantly dilute the overall effect as these patients have a preserved peakVO₂, 6MWT, and SF-36 score. Patients with a diminished TSAT however had a significantly worse peakVO₂, 6MWT, and SF-36, underscoring the importance of diminished TSAT in defining iron deficiency in PVD.

Figure 1.

Flowchart of study population and prevalence of iron deficiency according to novel transferrin saturation definition. Flowchart of patients in the study and the prevalence of iron deficiency by novel transferrin saturation definition. PVD, pulmonary vascular disease; TSAT, transferrin saturation; PH, pulmonary hypertension.

Figure 2.

Limitation of current guideline definition of iron deficiency in pulmonary vascular disease. Impact of the components of the heart failure definition of iron deficiency on peak oxygen consumption (panel A), 6-min walk testing (panel B), and 36-Item Short Form Survey physical component summary score (panel C). Patients with iron deficiency based on isolated hypoferritinemia (ferritin <100 ng/mL but transferrin saturation >20%) had similar peak oxygen consumption as patients without iron deficiency and better 6-min walk testing and 36-Item Short Form Survey, while patients with a low transferrin saturation component in the definition had significantly worse peak oxygen consumption, 6-min walk testing, and 36-Item Short Form Survey. P-values are from unadjusted t-test analysis. peakVO₂, peak oxygen consumption; TSAT, transferrin saturation; 6MWT, 6-min walk testing; SF-36, 36-Item Short Form Survey.

Secondly, we sought to better define iron deficiency. Supplementary data online, Figure S1 illustrates 3D Youden’s index plots for the different iron parameters in relation to different endpoints (peakVO₂, 6MWT, and HRQoL), illustrating the cutoff with the highest sensitivity and specificity for iron, TSAT, and ferritin for detection of a lower peakVO₂, 6MWT, and SF-36 score. The cut-point of TSAT <21% or iron <14 μmol/L best identified patients with a lower peakVO₂, 6MWT, and HRQoL with relative similar Youden’s index values, while ferritin had much lower values indicating less discriminating ability. Assessing the absolute between-group differences in peakVO₂, 6MWT, and HRQoL, the definition of TSAT <21% best identified a cohort of PVD patients with the largest degree of functional limitation (see Supplementary data online, Table S4) and also had the highest Youden’s index value in comparison to the iron-based definition. Therefore, a TSAT <21% was identified as the best cutoff and used in the remainder of the manuscript to define iron deficiency. Bootstrapping analysis provided internally confirmation of this cut-point (see Supplementary data online, Table S5). Of the patients with iron deficiency according to the heart failure definition, only 74.7% had also iron deficiency according to the TSAT <21% definition (see Supplementary data online, Table S6).

Prevalence of iron deficiency using transferrin saturation in the pulmonary vascular disease spectrum and controls

Figure 1 illustrates a flowchart of the overall study cohort (after exclusion of WSPH group 5) and the prevalence of iron deficiency defined by TSAT <21% in healthy individuals (n = 90), PVD comparators (n = 335), and PH patients (n = 724) with the distribution of iron values shown in Supplementary data online, Figure S2. Supplementary data online, Figure S3 shows similar findings for iron <14 µmol/L. There was a progressive increase in the proportion of patients suffering from iron deficiency from healthy controls to PVD comparators to PH patients. Table 1 illustrates baseline features of patients with and without iron deficiency across the PVD spectrum. Patients with PH and iron deficiency were more often women, had lower diastolic blood pressure, higher heart rate, lower sodium levels, higher N-terminal pro-B-type natriuretic peptide (NT-proBNP), higher C-reactive protein (in addition to lower iron parameter levels and hemoglobin), and more often used prostanoids. Supplementary data online, Table S7 illustrates the correlation between iron parameter and C-reactive protein showing a weak but statistically significant correlation. Pulmonary vascular disease comparator patients with iron deficiency were mostly female and also had a lower diastolic blood pressure, were more often black, and used phosphodiesterase 5 inhibitors more frequently.

Table 1.

Baseline features of patients with or without iron deficiency across the PVD spectrum

Parameter	Pulmonary hypertension (n = 693)			PVD comparator (n = 335)
	TSAT ≥21 (n = 266)	TSAT <21 (n = 427)	P-value	TSAT ≥21 (n = 176)	TSAT <21 (n = 159)	P-value
Demographics and clinical features
Age (years)	60 (47–70)	60 (49–69)	0.80	63 (56–70)	62 (52–69)	0.20
Female	147 (55%)	296 (69%)	<0.001	147 (55%)	296 (69%)	0.015
BMI (kg/m2)	28.8 (24.4–28.8)	28.8 (24.0–35.7)	0.29	29.9 (25.3–34.3)	29.4 (25.6–36.2)	0.086
SBP (mmHg)	127 (112–144)	122 (109–140)	0.074	134 (118–149)	129 (113–146)	0.34
DBP (mmHg)	74 (67–84)	71 (64–80)	<0.001	76 (68–84)	72 (65–82)	0.030
Heart rate (b.p.m.)	71 (64–80)	76 (66–86)	<0.001	72 (65–82)	69 (61–79)	0.37
Race			0.098			0.010
Black	25 (9.8%)	60 (14.3%)		7 (4.0%)	17 (11.0%)
White	216 (84.4%)	325 (77.6%)		164 (94.3%)	130 (83.9%)
Other	15 (5.9%)	34 (8.1%)		3 (1.7%)	8 (5.2%)
Laboratory features
eGFR (mL/min/1.73 m2)	76.5 (59.7–91.9)	73.0 (54.9–91.4)	0.33	77.0 (62.8–91.8)	73.2 (55.5–92.2)	0.11
Hemoglobin (g/dL)	14.3 (13.1–15.5)	13.1 (11.8–14.5)	<0.001	13.9 (12.9–14.8)	12.5 (11.3–13.6)	<0.001
Anemiaa	52 (20%)	182 (44%)	<0.001	35 (21%)	84 (54%)	<0.001
Sodium (mmol/L)	140 (139–142)	140 (138–141)	0.003	140 (139–142)	140 (138–142)	0.15
Iron (µmol/L)	18.0 (15.4–21.0)	9.1 (7.3–11.0)	<0.001	17.0 (15.0–19.0)	10 (8.0–12.0)	<0.001
Ferritin (ng/mL)	116 (66–225)	56 (27–116)	<0.001	125 (68–240)	55 (26–115)	<0.001
TSAT %	27.3 (23.9–32.6)	13.7 (10.0–17.3)		26.4 (23.4–32.4)	14.8 (10.9–17.7)
NT-proBNP (pg/mL)	242 (86–787)	463 (126–1600)	<0.001	118 (52–367)	145 (68–627)	0.040
C-reactive protein, mg/dL	2.6 (1.1–5.2)	4.3 (2.1–10.4)	<0.001	1.7 (0.9–4.1)	4.5 (2.0–10.3)	<0.001
WSPH groups
Group 1	129 (49%)	221 (49.4%)	0.189	35 (19.9%)	23 (14.5%)	0.597
Group 2	38 (14.3%)	93 (21.8%)		70 (39.8%)	70 (44.0%)
Group 3	71 (26.7%)	95 (22.2%)		58 (33.0%)	58 (36.5%)
Group 4	28 (10.5%)	28 (6.6%)		13 (7.4%)	8 (5.0%)
Medications
On PH medication	137 (51.9%)	252 (59.3%)	0.057	7 (4.0%)	13 (8.2%)	0.10
PDE5i	107 (40.5%)	195 (45.9%)	0.17	4 (2.3%)	11 (7.0%)	0.039
ERA	71 (26.9%)	143 (33.6%)	0.063	3 (1.7%)	4 (2.5%)	0.71
Prostanoids	48 (18.2%)	120 (28.2%)	0.003	0 (0.00%)	2 (1.3%)	0.22
sGC stimulators	15 (5.7%)	24 (5.6%)	0.98	1 (0.57%)	0 (0.00%)	0.99
On CCB meds for PH	11 (4.2%)	9 (2.1%)	0.12	0 (0.00%)	0 (0.00%)	n/a

PH, pulmonary hypertension; PVD, pulmonary vascular disease; BMI, body mass index; SBP, systolic blood pressure; DBP, diastolic blood pressure; AF, atrial fibrillation; eGFR, estimated glomerular filtration rate; TSAT, transferrin saturation; NT-proBNP, N-terminal pro-B-type natriuretic peptide; WSPH, World Symposium on Pulmonary Hypertension; PDE5i, phosphodiesterase 5 inhibitor; ERA, endothelin receptor antagonist; sGC, soluble guanylate cyclase; CCB, calcium channel blockers; n/a, not applicable.

World Health Organization definition.

Association of transferrin saturation–defined iron deficiency with exercise capacity and functional status

Figure 3 illustrates the differences in exercise capacity measured by peakVO₂ and 6MWT for the overall PVD spectrum, PH patients, and PVD comparators, after adjusting for age, sex, and hemoglobin levels. Patients with iron deficiency had a significant lower peakVO₂ in the PVD spectrum, PH patients, and PVD comparators. Supplementary data online, Figure S4 shows similar findings for iron <14 µmol/L, but of somewhat less effect as TSAT <21%. During cardiopulmonary exercise testing, PH patients with iron deficiency had a significantly lower oxygen pulse (TSAT <20%: 7.4 ± 2.4 mL/beat vs. TSAT >20%: 8.9 ± 3.4, P < 0.001), but similar respiratory exchange ratio (TSAT <20%: 1.1 ± 0.1 vs. TSAT >20%: 1.1 ± 0.2, P = 0.337) and V_E/V_CO2 slope (TSAT <20%: 40 ± 10 vs. TSAT >20%: 40 ± 11). Figure 3 additionally illustrates the difference in the SF-36 physical component score in patients with or without iron deficiency. After adjusting for age, sex, and hemoglobin, iron deficiency was associated with lower 6MWT and SF-36 physical component score in patients across the PVD spectrum and in PH patients, but not in the PVD comparators.

Figure 3.

Impact of iron deficiency on exercise capacity. Bar charts illustrating differences between iron-deficient and non-iron–deficient patients across the pulmonary vascular disease spectrum. Error bars indicate 95% confidence interval. P-values are from an analysis of covariance model adjusted for age, sex, and hemoglobin. Blue indicates transferrin saturation ≥21%, and red indicates transferrin saturation <21%. PVD, pulmonary vascular disease; PH, pulmonary hypertension; peakVO₂, peak oxygen consumption; 6MWT, 6-min walk testing; SF-36, 36-Item Short Form Survey.

Impact of anemia and the World Symposium on Pulmonary Hypertension group

In the PVD spectrum, a total of 353 (3.4.3%) patients had anemia. In patients without baseline anemia, the presence of iron deficiency was also associated with a low peakVO₂, 6MWT, and SF-36 (see Supplementary data online, Figure S5), and iron deficiency without anemia was associated with an equal reduction in peakVO₂, 6MWT, and SF-36 as anemic patients without iron deficiency, while patients with anemia and iron deficiency fared numerically the worse. Figure 4 illustrates the impact of iron deficiency in three separate forest plots for peakVO₂, 6MWT, and the SF-36 physical component score for PH subgroups WSPH Groups 1–4. No statistically significant interaction was observed in the relationship between iron deficiency and peakVO₂ and 6MWT among the different WSPH Groups 1–4. For SF-36, the P-value for interaction was significant, which was driven by Group 3 patients, who exhibited less functional limitation in the presence of iron deficiency. For patients in Groups 1 and 4, iron deficiency was consistently associated with worse peakVO₂, 6MWT, and SF-36. Supplementary data online, Figure S6 shows similar analysis for iron <14 µmol/L.

Figure 4.

Forrest plot of impact of iron deficiency according to the World Symposium on Pulmonary Hypertension group. Subgroup analysis from a generalized linear regression model with an interaction term of iron deficiency and the World Symposium on Pulmonary Hypertension groups. Blue indicates transferrin saturation ≥21%, and red indicates transferrin saturation <21%. WSPH, World Symposium on Pulmonary Hypertension; peakVO₂, peak oxygen consumption; 6MWT, 6-min walk testing; SF-36, 36-Item Short Form Survey.

Association between cardiac function and structure and iron deficiency

Of the 1028 patients in the PVD spectrum, a total of 613 patients (59.6%), including 425 PH patients (61.3%) and 188 PVD comparators (56.1%), underwent CMR imaging. Table 2 illustrates the results of an ANCOVA model showing the differences in right and left ventricular function and structure in patients with or without iron deficiency after adjusting for sex (given sex-based difference in chamber volumes). In the PVD spectrum, patients with iron deficiency had a similar left ventricular function and structure; however, patients with iron deficiency had more left atrial and RV remodeling, including larger RV volumes and mass and lower RV ejection fraction with larger right atrial volumes, but these differences were not apparent in the smaller cohort of patients with PH alone. No significant differences were present in RV remodeling in PVD comparator patients, while these patients did have signs of a more dilated left atrium with reduced global longitudinal strain.

Table 2.

Difference in cardiac structure and function measured by CMR in patients with and without iron deficiency

Parameter	PVD spectrum (n = 613)			PH patients (n = 425)			PVD comparators (n = 188)
	No iron deficiency	Iron deficiency	P-value	No iron deficiency	Iron deficiency	P-value	No iron deficiency	Iron deficiency	P-value
Left heart parameters a
LVEDV, mL	128 (124–133)	133 (129–138)	0.114	127 (121–133)	131 (126–136)	0.309	129 (121–138)	139 (131–148)	0.119
LVESV, mL	58 (55–59)	61 (58–64)	0.052	56 (52–60)	60 (57–63)	0.120	56 (51–62)	64 (58–70)	0.051
LVEF, %	56 (56–58)	55 (55–57)	0.118	56 (55–58)	56 (55–57)	0.577	58 (56–59)	55 (53–57)	0.056
LV GLS	−12 (−12 to −11)	−12 (−12 to −11)	0.615	−12 (−12 to −11)	−12 (12 to −11)	0.937	−12 (−13 to −12)	−12 (−13 to −11)	0.617
LV mass, g	88 (84–92)	91 (88–94)	0.054	87 (83–91)	91 (87–94)	0.191	86 (81–91)	93 (88–98)	0.060
LV SV, mL	72 (69–74)	72 (70–75)	0.946	71 (68–75)	71 (69–74)	0.911	73 (69–78)	75 (71–80)	0.557
LA EDV, mL	34 (30–38)	40 (36–43)	0.023	33 (29–38)	38 (35–42)	0.099	35 (28–41)	44 (37–51)	0.074
LA ESV, mL	60 (55–64)	67 (63–70)	0.023	58 (53–64)	64 (59–68)	0.143	62 (54–70)	75 (67–83)	0.027
LA EF, %	57 (56–58)	56 (55–57)	0.118	45 (43–48)	43 (41–45)	0.103	48 (45–51)	46 (43–50)	0.533
Right heart parameters a
RVEDV, mL	166 (158–174)	184 (177–191)	0.002	182 (171–193)	198 (189–206)	0.032	139 (129–148)	148 (138–158)	0.189
RVESV, mL	96 (89–103)	112 (105–118)	0.002	113 (103–123)	125 (117–133)	0.048	67 (60–74)	75 (67–82)	0.142
RVEF, %	45 (43–46)	43 (41–44)	0.030	40 (39–42)	40 (38–41)	0.498	53 (51–54)	51 (49–53)	0.254
RV GLS	−17 (−18 to−17)	−17 (−18 to −17)	0.254	−17 (−17 to −16)	−16 (−17 to −15)	0.335	−20 (−21 to −19)	−20 (−21 to −19)	0.641
RV mass, g	33 (31–35)	37 (35–38)	0.003	36 (34–39)	40 (38–42)	0.026	27 (25–29)	28 (26–30)	0.522
RV SV, mL	70 (68–73)	72 (70–74)	0.235	69 (66–72)	72 (69–75)	0.199	23 (20–26)	24 (21–27)	0.664
RA EDV, mL	46 (42–51)	55 (51–60)	0.004	50 (44–56)	61 (56–66)	0.006	40 (33–47)	41 (34–49)	0.790
RA ESV, mL	71 (65–76)	82 (78–87)	0.001	75 (68–81)	88 (83–94)	0.002	63 (55–72)	65 (57–74)	0.699
RA EF, %	37 (36–39)	36 (34–38)	0.276	36 (34–39)	34 (33–36)	0.181	53 (51–54)	51 (49–53)	0.254

PH, pulmonary hypertension; PVD, pulmonary vascular disease; LV, left ventricle; EDV, end-diastolic volume; ESV, end-systolic volume; EF, ejection fraction; GLS, global longitudinal strain; SV, stroke volume; LA, left atrium; RV, right ventricle; RA, right atrium.

Results are from an ANCOVA model adjusted for female sex as more females were prevalent in the iron deficiency group and females have different cutoffs for chamber volumes. Results are presented as least square mean and 95% CI.

Association between hemodynamics and iron deficiency

Table 3 illustrates the right heart catheterization results for patients across the PVD spectrum, PH patients, and PVD comparators. In the overall PVD spectrum, patients with iron deficiency had evidence of higher filling pressures including right atrial pressures (RAP), systolic/mean/diastolic pulmonary pressures, and pulmonary capillary wedge pressure. Additionally, patients had a higher heart rate and lower resistance/compliance (RC) time constant; however, PVR and cardiac output were similar. Similar results were observed in PH patients and PVD comparators. As the RC time constant is biologically dependent on pulmonary capillary wedge pressure, heart rate, and body surface area, an adjusted ANCOVA analysis was performed for the overall PVD spectrum, PH patients, and PVD comparators, with no difference in RC time constant between patients with or without iron deficiency after adjusting for these covariates (P > 0.593 for all).

Table 3.

Hemodynamic measurements during resting right heart catheterization

Parameter	PVD spectrum (n = 1011)			PH patients (n = 686)			PVD comparators (n = 325)
	No iron deficiency	Iron deficiency	P-value	No iron deficiency	Iron deficiency	P-value	No iron deficiency	Iron deficiency	P-value
Heart rate, b.p.m.	72 ± 14	75 ± 14	<0.001	73 ± 13	77 ± 15	0.001	71 ± 14	72 ± 11	0.372
RAP, mmHg	7 ± 5	8 ± 5	<0.001	8 ± 5	9 ± 5	0.034	6 ± 4	7 ± 4	0.046
Systolic PAP, mmHg	50 ± 24	57 ± 24	<0.001	63 ± 21	67 ± 21	0.050	31 ± 10	34 ± 11	0.008
Mean PAP, mmHg	32 ± 15	36 ± 15	<0.001	40 ± 14	42 ± 13	0.110	19 ± 6	22 ± 7	0.001
Diastolic PAP, mmHg	21 ± 11	24 ± 11	<0.001	26 ± 10	27 ± 10	0.228	13 ± 5	15 ± 6	0.009
Pulse pressure, mmHg	29 ± 15	34 ± 16	<0.001	37 ± 14	39 ± 15	0.042	18 ± 7	19 ± 7	0.072
PCWP mean, mmHg	12 ± 6	13 ± 7	<0.001	12 ± 6	13 ± 7	0.010	11 ± 5	12 ± 6	0.009
PCWP a-wave, mmHg	14 ± 6	15 ± 7	0.002	14 ± 6	15 ± 7	0.025	13 ± 6	15 ± 6	0.052
PCWP v-wave, mmHg	14 ± 8	17 ± 9	<0.001	15 ± 8	17 ± 9	0.001	14 ± 7	16 ± 9	0.013
CO, L/min	5.2 ± 1.6	5.3 ± 1.8	0.606	5.0 ± 1.6	5.1 ± 1.6	0.535	5.5 ± 1.7	5.7 ± 1.9	0.586
CI, L/min/m²	2.7 ± 0.7	2.7 ± 0.8	0.690	2.6 ± 0.8	2.7 ± 0.8	0.648	2.7 ± 0.7	2.7 ± 0.8	0.811
PVR, WU	4.3 (2.7–6.9)	4.1 (2.5–7.3)	0.698	5.0 (3.6–8.0)	5.0 (3.2–7.9)	0.633	1.6 (1.1–2.3	1.9 (1.1–2.5)	0.118
RC time constant, s	0.57 ± 0.19	0.51 ± 0.18	<0.001	0.61 ± 0.16	0.57 ± 0.16	0.001	0.43 ± 0.18	0.41 ± 0.16	0.364

RC time constant is also known as tau.

RAP, right atrial pressure; PAP, pulmonary arterial pressure; PCWP, pulmonary capillary wedge pressure; CO, cardiac output; CI, cardiac index; PVR, pulmonary vascular resistance; RC, resistance/compliance.

Association between iron deficiency and outcome

During a median follow-up of 36^24–37 months, a total of 162 (16%) patients in the overall PVD spectrum died. After adjusting for age, sex, hemoglobin, and NT-proBNP, patients with TSAT ≤21% displayed a 52% higher risk of all-cause mortality (hazard ratio 1.63, 95% confidence interval 1.13–2.34; P = 0.009).

Discussion

This comprehensive analysis from the large, multicenter PVDOMICS network provides novel and important insights into the role of iron deficiency in PVD. Here, we observed that (i) the 2022 ESC/ERS guideline–endorsed definition of iron deficiency does not identify patients with altered functional status or exercise capacity; (ii) a definition based solely on TSAT (<21%) best identifies patients with reduced exercise capacity and functional status; (iii) iron deficiency is present in up to 62% of patients with PH; (iv) the association between iron deficiency and reduced exercise capacity and functional status is independent of the presence of anemia and consistent across WSPH groups (with exception of SF-36 and Group 3 patients); (v) iron deficiency is associated with more pronounced right heart remodeling; and (vi) patients with iron deficiency did not have more elevated PVR but did exhibit slightly higher filling pressures (Structured Graphical Abstract).

Importance of optimally defining iron deficiency

Multiple randomized placebo-controlled trials have shown that treatment of iron deficiency with intravenous iron leads to improvement in functional status, exercise capacity, cardiac reverse remodeling, and risk for heart failure readmissions in patients with heart failure.^{2,5,8,28–31} The 2022 ESC/ERS PH guidelines adopted the heart failure definition of iron deficiency, in an effort to enhance recognition of this important comorbidity and align with heart failure guidelines.^11,32 While such alignment of definitions makes sense as patients with PH might have left-sided heart disease or develop right-sided heart failure, this definition is not formally validated in the field of PH, and most observational studies in PH used different definitions to define iron deficiency (see Supplementary data online, Table S8).¹¹ Proper validation of the definition of iron deficiency is important for prognostication but also for providing a working definition that can be used to identify suitable patients for inclusion in randomized controlled trials with agents alleviating iron deficiency. Indeed, the 2022 ESC/ERS PH guidelines recognize that there is insufficient data from randomized trials to support the use of intravenous iron (especially in non-anemic patients), underscoring the need for additional randomized trials.¹¹ In that aspect, our analysis is of major importance. Importantly, the current definition of iron deficiency in the 2022 ESC/ERS guidelines does not identify patients with worse functional alterations. Therefore, it is questionable that using this definition in trials will lead to significant improvements in functional status and exercise capacity. The finding that a TSAT <21% best identifies these patients with worse functional status and exercise capacity is also consistent with more recent studies in the field of heart failure, confirming that functional abnormalities and adverse cardiac remodeling are worse in patients with a low TSAT.³³ Patients with a low TSAT show the most convincing treatment effect with intravenous iron in heart failure studies in terms of functional improvement and cardiac reverse remodeling.^5,26 Interestingly, validation studies in heart failure have also suggested that a definition based on TSAT alone might best identify iron deficiency, with a TSAT cut-point of 19.8% being proposed, which is very close to our current definition in PH.²⁷ Our novel definition of iron deficiency is analyzed in a large cohort of PH patients and therefore better calibrated for PH patients. Importantly, we show that iron deficiency is associated with worse functional status and exercise capacity even after correcting for age, sex, and hemoglobin. Indeed, trials in the field of heart failure have shown beneficial effects of intravenous iron on functional status and exercise capacity even in the absence of anemia. Currently, the ESC/ERS PH guidelines do not give a recommendation for intravenous iron in patients with iron deficiency without anemia, given the absence of adequately powered trials in this field. Indeed, assuming a treatment effect of intravenous iron of around 20 m improvement in 6MWT (with a 50 m standard deviation) and α=0.05% and 90% power, around 262 patients should be randomized assuming no dropout. To date, no such trials exist in the field of PH with iron therapies.

Clinical associations with cardiac and pulmonary vascular remodeling

An additional strength of the PVDOMICS cohort is that it prospectively assessed a large cohort of PVD patients, and the PVD comparator patients are reflective of the lower PVR and mPAP thresholds recently introduced in the 2022 ESC/ERS guidelines.¹¹ Furthermore, all patients underwent a protocol-driven analysis with imaging, exercise testing, and right heart catheterization with central core lab adjudication. As pre-clinical and clinical studies suggest that iron deficiency might induce cardiac remodeling and pulmonary vascular remodeling, we assessed the association between TSAT and cardiac and vascular remodeling. On CMR analysis, patients with iron deficiency had more RV remodeling and some signs of diminished RV function, but there were no differences in left ventricular function or structure. Much of the mortality and morbidity in PH is related to the development of maladaptive RV remodeling. Positron emission tomography analysis has shown that such remodeling is associated with enhanced glucose flux.¹⁰ As iron importantly regulates cardiac energetics (both the electron transport chain and the tricarboxylic acid cycle enzymes), it begs the question if iron deficiency directly contributes to the failing RV in PH.³ Interestingly, patients with milder PH (PVD comparators) had less iron deficiency and were still without signs of maladaptive RV remodeling. The interest in the relation between iron and RV remodeling is further strengthened by two studies in the field of heart failure (WSPH Group 2 PH) showing that treatment with ferric carboxymaltose induced cardiac reverse remodeling measured by either CMR (RV ejection fraction or global longitudinal strain) or echocardiography [RV fractional area change, tricuspid annular plane systolic excursion (TAPSE), or pulmonary artery systolic pressure/TAPSE ratio], and notably, these variables are well-known prognosticators in PH.^8,34,35 Indeed, adjusting for baseline NT-proBNP, age, sex, and hemoglobin, patients with a TSAT <21% had higher risk of all-cause mortality. Looking at the right heart catheterization results, iron deficiency is not associated with more severe pulmonary vascular remodeling measured by PVR. This is in contrast to some pre-clinical studies suggesting that iron deficiency might be associated with progressive pulmonary vascular remodeling.^36,37

Limitations

Several limitations need to be acknowledged. First, other informative laboratory values such as soluble transferrin receptors or hepcidin were not available. Second, the absence of significant interaction in the analysis for WSPH might also be related to the often underpowered nature of interaction tests, which do not always detect a true biological difference (Type 2 error). Third, validation against the gold standard of bone marrow iron staining was not available. Finally, we describe a cut-point for TSAT to define iron deficiency based on the PVDOMICS cohort, but external validation of this cut-point in a separate cohort seems desirable. Nevertheless, our internal validation through bootstrapping identifies a similar cut-point for TSAT <21%. Dedicated randomized controlled trials using our proposed definition for iron deficiency will be helpful to better determine the role of iron therapies in improving functional status, exercise capacity, cardiac structure, and outcome in PH.

Conclusion

Iron deficiency defined as a TSAT ≤21% is common in patients across the PVD spectrum and is associated with worse functional status, exercise capacity, right heart remodeling and outcome, irrespective of the presence of anemia.

Author contributions

Pieter Martens (Conceptualization [equal], Formal analysis [lead], Methodology [lead], Writing—original draft [lead]), Erika Berman Rosenzweig (Data curation [supporting], Funding acquisition [supporting], Writing—review & editing [supporting]), Franz P. Rischard (Data curation [supporting], Funding acquisition [supporting], Writing—review & editing [supporting]), Margaret M. Park (Data curation [supporting], Writing—review & editing [supporting]), Deborah H. Kwon (Data curation [supporting], Writing—review & editing [supporting]), Miriam Jacob (Data curation [supporting], Writing—review & editing [supporting], Evelyn M. Horn (Data curation [supporting], Funding acquisition [supporting], Writing—review & editing [supporting]), Nicholas S. Hill (Data curation [supporting], Writing—review & editing [supporting]), Anna R. Hemnes (Data curation [supporting], Funding acquisition [supporting], Writing—review & editing [supporting]), Gabriele Grunig (Writing—review & editing [supporting]), J. Emanuel Finet (Data curation [supporting], Writing—review & editing [supporting]), Samar Farha (Writing—review & editing [supporting]), Serpil Erzurum (Data curation [supporting], Funding acquisition [equal], Writing—review & editing [supporting]), Barry Borlaug (Data curation [supporting], Writing—review & editing [supporting]), Brett Larive (Data curation [supporting], Formal analysis [supporting], Writing—review & editing [supporting]), Shilin Yu (Formal analysis [supporting], Writing—review & editing [supporting]), Jennifer Wilcox (Data curation [supporting], Writing—review & editing [supporting]), and Wai Hong Wilson Tang (Conceptualization [equal], Data curation [lead], Funding acquisition [supporting], Supervision [lead], Writing—original draft [supporting], Writing—review & editing [supporting])

Supplementary data

Supplementary data is available at European Heart Journal online.

Pre-registered clinical trial number

Clinical trial registration: ClinicalTrials.gov NCT02980887

Ethical approval

All patients provided written informed consent.

Data availability

Anonymized data can be made available upon appropriate data sharing requests through formal application to the PVDOMICS steering committee, taking into account the local legal and ethical framework around anonymized data sharing.

Funding

The study received grants U01 HL125218 (PI: E.B.R.), U01 HL125205 (PI: R.P.F.), U01 HL125212 (PI: A.R.H.), U01 HL125208 (PI: F.P.R.), U01 HL125215 (PI: J.A.L.), and U01 HL125177 (PI: G.J.B.) and grants from the Pulmonary Hypertension Association. P.T. is supported by a grant from the Belgian American Educational Foundation (BAEF) and the Frans van de Werf Fund.