Increased airway iron parameters and risk for exacerbation in COPD: an analysis from SPIROMICS

Abstract

Levels of iron and iron-related proteins including ferritin are higher in the lung tissue and lavage fluid of individuals with chronic obstructive pulmonary disease (COPD), when compared to healthy controls. Whether more iron in the extracellular milieu of the lung associates with distinct clinical phenotypes of COPD, including increased exacerbation susceptibility, is unknown. We measured iron and ferritin levels in the bronchoalveolar lavage fluid (BALF) of participants enrolled in the SubPopulations and InteRmediate Outcome Measures In COPD (SPIROMICS) bronchoscopy sub-study (n = 195). BALF Iron parameters were compared to systemic markers of iron availability and tested for association with FEV₁ % predicted and exacerbation frequency. Exacerbations were modelled using a zero-inflated negative binomial model using age, sex, smoking, and FEV₁ % predicted as clinical covariates. BALF iron and ferritin were higher in participants with COPD and in smokers without COPD when compared to non-smoker control participants but did not correlate with systemic iron markers. BALF ferritin and iron were elevated in participants who had COPD exacerbations, with a 2-fold increase in BALF ferritin and iron conveying a 24% and 2-fold increase in exacerbation risk, respectively. Similar associations were not observed with plasma ferritin. Increased airway iron levels may be representative of a distinct pathobiological phenomenon that results in more frequent COPD exacerbation events, contributing to disease progression in these individuals.

Introduction

Chronic obstructive pulmonary disease (COPD) is a leading cause of mortality and morbidity worldwide, resulting in almost 3 million deaths globally in 2016 and the loss of over 47 million potential life-years¹. A significant portion of this burden of disease presents as exacerbation events, episodic surges of respiratory symptoms accompanied by a more rapid decline in lung function and higher mortality^2,3. These exacerbations are largely attributed to recurrent acute infection, with host factors also playing a fundamental role⁴. Identifying and understanding host susceptibility risk factors that contribute to repeated exacerbations is imperative to understanding and treating COPD.

Although the overwhelming risk factor in the industrialized world for COPD development is tobacco smoking, genome wide association studies suggest a pathogenic role for abnormal iron homeostasis⁵. We recently demonstrated that a major iron metabolism protein, iron regulatory protein 2 (IRP2), drives lung inflammation and injury in a murine model of COPD^5,6. In the lung, iron is found in both unbound and protein-bound forms, and several of the most abundant proteins in lung tissue and bronchoalveolar lavage fluid (BALF) bind to and regulate iron^7,8,9. One such protein is ferritin, an octahedral polymeric shell composed of light chain (FTL) and heavy chain (FTH) subunits that stores ferric (Fe³⁺) iron atoms in a soluble, non-toxic form¹⁰. Previous studies have demonstrated release of ferritin from iron-loaded alveolar macrophages (AMs) in smokers, and AM ferritin mRNA levels are increased in active smokers and correlate with airflow limitation in COPD patients^11,12. Furthermore, total levels of non-heme iron and of other iron-binding molecules including lipocalin-2 and lactoferrin, are increased in lung tissue, sputum, BALF, and AMs of COPD patients, relative to non-smokers^{9,11,12,13,14,15,16,17,18,19,20,21}. Conversely, there is also ample evidence for iron deficiency in COPD, and anaemia in COPD is associated with worse patient outcomes, including mortality^22,23. The biological relevance of such observations remains to be elucidated; however, these data strongly support a local iron overload signature in the extracellular milieu of the lung in COPD that is distinctive to systemic iron handling, which is intriguing as mainstream cigarette smoke contains little iron^24,61.

Clinical data was collected at the baseline study visit and in follow-up visits as previously described²⁷. Peripheral blood was collected as part of the baseline visit, and plasma biomarkers were measured using a Luminex-based multiplex assay; relevant to this study, plasma ferritin was shown to be equivalent to serum ferritin using this assay method⁶². Exacerbations were defined as health care utilization events (office visit, hospital admission, or emergency department visit for a respiratory flare-up) that were treated with antibiotics, systemic corticosteroids, or both. Exacerbation history was prospectively collected every 3 months for up to 5 years using a structured questionnaire^2,60.

A subgroup of subjects (n = 215) with post-bronchodilator FEV₁ > 30% predicted and without an exacerbation in the prior six weeks were further enrolled in the bronchoscopy sub-study, in which on the first of two visits, sputum induction was performed as previously described⁶³. On the second visit, post-bronchodilator FEV₁ was measured and only subjects with an FEV₁ > 30% predicted were allowed to participate in the bronchoscopy portion of the study. BAL was performed in the right middle lobe and lingula by instilling two aliquots of 40 mL and one aliquot of 50 mL per lobe (260 mL total volume), after excluding an initial airway wash sample. Unfiltered BALF fluid was collected into a sterilized beaker or in multiple 50 mL conical tubes on ice, then centrifuged at 300 x g for 5 min and the supernatant aliquoted into 1 mL aliquots for storage at −80 °C, representing one BALF sample per patient as described previously^63,64. Because no research plasma samples were obtained in the bronchoscopy sub-study, the plasma biomarkers, including ferritin, in this analysis were measured from the baseline visit samples. For participants in the bronchoscopy sub-study, exacerbation events were analysed both relative to the baseline visit and to the bronchoscopy visit (0-14 months after baseline visit).

To replicate the study findings in an independent cohort, never smokers (n = 20), healthy smokers (n = 21) with normal lung function and individuals with COPD (n = 18), recruited by the Department of Genetic Medicine, Weill Cornell Medical College underwent bronchoscopy with BALF isolated as described above (see Supplementary Information and Supplemental Table 1).

Ferritin measurement and normalization

BALF ferritin was quantified by ELISA using the Abcam Human Ferritin ELISA Kit (Cat#ab200018), which detects both ferritin heavy and light chain. BALF ferritin was normalized to total protein, measured using the Thermo Scientific Pierce BCA Protein Assay Kit (Cat#23225).

Total iron measurements

After centrifugation (1000 × g for 5 mins), 60 μL of BALF was digested with 40 μL of 50% Nitric Acid (in distilled H₂O) containing a final concentration of 0.1% digitonin for 2 hours at 60 °C. Total iron, including both bound and unbound forms, was measured in triplicate in 20 μL of digested fractions using a graphite furnace atomic absorption spectrophotometer (GFAAS, Perkin Elmer PinAAcle 900z), comparing unknown values to a standard curve of known concentrations of iron (1000 PPM in 2% Nitric Acid).

Statistical analysis

Clinical characteristics of SPIROMICS participants enrolled in the bronchoscopy sub-study were compared to those of all SPIROMICS participants, summarized using means and standard deviations or counts and percentages as appropriate. BALF Ferritin and iron were analysed on the log 10 scale. Plasma ferritin, haemoglobin and CRP were measured as previously described⁶², and analysed on the log scale, accounting for site and batch effects. Associations between BALF ferritin and baseline characteristics were performed using Kruskal-Wallis tests for categorical variables, and Pearson correlations for continuous variables, and were unadjusted unless otherwise specified. Sensitivity analyses including all SPIROMICS participants were also performed to study associations in the overall cohort. COPD exacerbations were analyzed in three ways. First, participants were dichotomized into those with any exacerbations between their baseline visit and the end of study follow-up versus those without exacerbations in this timeframe, and ferritin levels were compared across groups. Second, participants were dichotomized into those with any exacerbations between the bronchoscopy visit and the end of the study follow-up, versus those without, and ferritin levels were compared across groups. Third, the rate of exacerbations per participant per year was estimated using a negative binomial zero-inflated model, with % FEV₁ predicted as the predictor in the binomial model. Ferritin associations were studied in models unadjusted as well as adjusted for age, sex, and smoking status at baseline or at the time of the bronchoscopy, as appropriate. Adjustment for study site prevented model convergence and was thus removed. Yearly exacerbation rate ratios, as well as 95% confidence intervals (CI) were estimated, and the unadjusted model was then used to plot predicted exacerbation rates in three years in a participant with a median FEV_1% predicted. BALF ferritin levels in a validation cohort (see Supplemental Table 1) were similarly analysed on a log 10 scale, and compared across non-smokers, smokers without COPD, and participants with COPD.