The role and potential of imaging in COPD

Machine learning and deep learning predictive models for long-term prognosis in patients with chronic obstructive pulmonary disease: a systematic review and meta-analysis

BACKGROUND

Machine learning and deep learning models have been increasingly used to predict long-term disease progression in patients with chronic obstructive pulmonary disease (COPD). We aimed to summarise the performance of such prognostic models for COPD, compare their relative performances, and identify key research gaps.

METHODS

We conducted a systematic review and meta-analysis to compare the performance of machine learning and deep learning prognostic models and identify pathways for future research. We searched PubMed, Embase, the Cochrane Library, ProQuest, Scopus, and Web of Science from database inception to April 6, 2023, for studies in English using machine learning or deep learning to predict patient outcomes at least 6 months after initial clinical presentation in those with COPD. We included studies comprising human adults aged 18-90 years and allowed for any input modalities. We reported area under the receiver operator characteristic curve (AUC) with 95% CI for predictions of mortality, exacerbation, and decline in forced expiratory volume in 1 s (FEV1). We reported the degree of interstudy heterogeneity using Cochran's Q test (significant heterogeneity was defined as p≤0·10 or I2>50%). Reporting quality was assessed using the TRIPOD checklist and a risk-of-bias assessment was done using the PROBAST checklist. This study was registered with PROSPERO (CRD42022323052).

FINDINGS

We identified 3620 studies in the initial search. 18 studies were eligible, and, of these, 12 used conventional machine learning and six used deep learning models. Seven models analysed exacerbation risk, with only six reporting AUC and 95% CI on internal validation datasets (pooled AUC 0·77 [95% CI 0·69-0·85]) and there was significant heterogeneity (I2 97%, p<0·0001). 11 models analysed mortality risk, with only six reporting AUC and 95% CI on internal validation datasets (pooled AUC 0·77 [95% CI 0·74-0·80]) with significant degrees of heterogeneity (I2 60%, p=0·027). Two studies assessed decline in lung function and were unable to be pooled. Machine learning and deep learning models did not show significant improvement over pre-existing disease severity scores in predicting exacerbations (p=0·24). Three studies directly compared machine learning models against pre-existing severity scores for predicting mortality and pooled performance did not differ (p=0·57). Of the five studies that performed external validation, performance was worse than or equal to regression models. Incorrect handling of missing data, not reporting model uncertainty, and use of datasets that were too small relative to the number of predictive features included provided the largest risks of bias.

INTERPRETATION

There is limited evidence that conventional machine learning and deep learning prognostic models demonstrate superior performance to pre-existing disease severity scores. More rigorous adherence to reporting guidelines would reduce the risk of bias in future studies and aid study reproducibility.

FUNDING

None.

Aging-Related Findings of the Respiratory System in Chest Imaging: Pearls and Pitfalls

Purpose of Review

The purpose of this review is to describe the main features of the aging chest, studied through different imaging modalities.

Recent Findings

Aging-related changes of the respiratory system are inevitable. Therefore, it is mandatory to be familiar with the para-physiological changes that occurs, in order to avoid inappropriate interpretation of radiological findings that put patients at risk of over or undertreatment.

Summary

The role of the radiologist is fundamental in evaluating aging-related processes affecting the respiratory system and in distinguishing them from frank diseases.

PM2.5 induces pulmonary microvascular injury in COPD via METTL16-mediated m6A modification

Fine particulate matter (PM2.5) exposure is a significant cause of chronic obstructive pulmonary disease (COPD), but the detailed mechanisms involved in COPD remain unclear. In this study, we established PM2.5-induced COPD rat models and showed that PM2.5 induced pulmonary microvascular injury via accelerating vascular endothelial apoptosis, increasing vascular permeability, and reducing angiogenesis, thereby contributing to COPD development. Moreover, microvascular injury in COPD was validated by measurements of plasma endothelial microparticles (EMPs) and serum VEGF in COPD patients. We then performed m⁶A sequencing, which confirmed that altered N⁶-methyladenosine (m⁶A) modification was induced by PM2.5 exposure. The results of a series of experiments demonstrated that the expression of methyltransferase-like protein 16 (METTL16), an m⁶A regulator, was upregulated in PM2.5-induced COPD rats, while the expression of other regulators did not differ upon PM2.5-induction. To clarify the regulatory effect of METTL16-mediated m⁶A modification induced by PM2.5 on pulmonary microvascular injury, cell apoptosis, permeability, and tube formation, the m⁶A level in METTL16-knockdown pulmonary microvascular endothelial cells (PMVECs) was evaluated, and the target genes of METTL16 were identified from a set of the differentially expressed and m⁶A-methylated genes associated with vascular injury and containing predicted sites of METTL16 methylation. The results showed that Sulfatase 2 (Sulf2) and Cytohesin-1 (Cyth1) containing the predicted METTL16 methylation sites, exhibited higher m⁶A methylation and were downregulated after PM2.5 exposure. Further studies demonstrated that METTL16 may regulate Sulf2 expression via m⁶A modification and thereby contribute to PM2.5-induced microvascular injury. These findings not only provide a better understanding of the role played by m⁶A modification in PM2.5-induced microvascular injury, but also identify a new therapeutic target for COPD.

Abnormal pulmonary flow is associated with impaired right ventricular coupling in patients with COPD

Cor Pulmonale or right ventricular (RV) dysfunction due to pulmonary disease is an expected complication of COPD resulting primarily from increased afterload mediated by chronic alveolar hypoxemia and resulting hypoxic pulmonary vasoconstriction. Early detection of elevated RV afterload has been previously demonstrated by visualization of abnormal flow patterns in the proximal pulmonary arteries. Prior analysis of helicity in the pulmonary arteries in pulmonary hypertension patients has demonstrated a strong association between helicity and increased RV afterload. However, these flow hemodynamics have yet to be fully explored in patients with COPD. We hypothesized that patients with COPD will have abnormal pulmonary flow as evaluated by 4D-Flow MRI and associated with RV function and pulmonary arterial stiffness. Patients with COPD (n = 15) (65 years ± 6) and controls (n = 10) (58 years ± 9) underwent 4D-Flow MRI to calculate helicity. The helicity was calculated in the main pulmonary artery (MPA) and along the RV outflow tract (RVOT)-MPA axis. Main pulmonary arterial stiffness was measured using the relative area change (RAC). We found COPD patients had decreased helicity relative to healthy controls in the MPA (19.4 ± 7.8vs 32.8 ± 15.9, P = 0.007) and reduced helicity along the RVOT-MPA axis (33.2 ± 9.0 vs 43.5 ± 8.3, P = 0.010). Our investigation indicates a strong association between helicity along the MPA-RV outflow tract axis and RV function and suggests that 4D-Flow MRI might be a sensitive tool in evaluating RV-pulmonary arterial coupling in COPD.

COPD: To Be or Not to Be, That is the Question

As our knowledge on the natural history of chronic obstructive pulmonary disease (COPD) progresses, a conceptual model simply based on an accelerated decline of lung function in adult life in response to smoking has become inadequate to capture the complexity of this disease, and increasing attention is being given to possible contributions from events or alterations of developmental processes that take place earlier in life. In addition, a remarkable heterogeneity has emerged among the pathobiological mechanisms that are involved in different phenotypes of COPD, suggesting that an effective disease management will require individualized treatment approaches largely based on the underlying biological mechanisms (endotypes). In this review, we will discuss the many faces of COPD from an epidemiological, pathobiological, and clinical standpoint and argue that airflow limitation encompasses a number of manifestations that are too diverse to be still clustered under the same diagnostic label.

Molecular Characteristics and Treatment of Endothelial Dysfunction in Patients with COPD: A Review Article

Patients with chronic obstructive pulmonary disease (COPD) show systemic consequences, such as chronic systemic inflammation leading to changes in the airway, airway penetrability, and endothelial function. Endothelial dysfunction is characterized by a list of alterations of endothelium towards reduced vasodilation, proinflammatory state, detachment and apoptosis of endothelial cells, and development of atherosclerosis. COPD-induced endothelial dysfunction is associated with elevated cardiovascular risk. The increment of physical activities such as pulmonary rehabilitation (PR) training have a significant effect on COPD, thus, PR can be an integrative part of COPD treatment. In this narrative review the focus is on the function of endothelial inflammatory mediators [cytokines, chemokines, and cellular proteases] and pulmonary endothelial cells and endothelial dysfunction in COPD as well as the effects of dysfunction of the endothelium may play in COPD-related pulmonary hypertension. The relationship between smoking and endothelial dysfunction is also discussed. The connection between different pulmonary rehabilitation programs, arterial stiffness and pulse wave velocity (PWV) is presented. Endothelial dysfunction is a significant prognostic factor of COPD, which can be characterized by PWV. We discuss future considerations, like training programs, as an important part of the treatment that has a favorable impact on the endothelial function.

Artificial intelligence in diagnosis of obstructive lung disease: current status and future potential

PURPOSE OF REVIEW

The application of artificial intelligence in the diagnosis of obstructive lung diseases is an exciting phenomenon. Artificial intelligence algorithms work by finding patterns in data obtained from diagnostic tests, which can be used to predict clinical outcomes or to detect obstructive phenotypes. The purpose of this review is to describe the latest trends and to discuss the future potential of artificial intelligence in the diagnosis of obstructive lung diseases.

RECENT FINDINGS

Machine learning has been successfully used in automated interpretation of pulmonary function tests for differential diagnosis of obstructive lung diseases. Deep learning models such as convolutional neural network are state-of-the art for obstructive pattern recognition in computed tomography. Machine learning has also been applied in other diagnostic approaches such as forced oscillation test, breath analysis, lung sound analysis and telemedicine with promising results in small-scale studies.

SUMMARY

Overall, the application of artificial intelligence has produced encouraging results in the diagnosis of obstructive lung diseases. However, large-scale studies are still required to validate current findings and to boost its adoption by the medical community.

COPD and Cardiovascular Disease

COPD is one of the major public health problems in people aged 40 years or above. It is currently the 4th leading cause of death in the world and projected to be the 3rd leading cause of death by 2020. COPD and cardiac comorbidities are frequently associated. They share common risk factors, pathophysiological processes, signs and symptoms, and act synergistically as negative prognostic factors. Cardiac disease includes a broad spectrum of entities with distinct pathophysiology, treatment and prognosis. From an epidemiological point of view, patients with COPD are particularly vulnerable to cardiac disease. Indeed, mortality due to cardiac disease in patients with moderate COPD is higher than mortality related to respiratory failure. Guidelines reinforce that the control of comorbidities in COPD has a clear benefit over the potential risk associated with the majority of the drugs utilized. On the other hand, the true survival benefits of aggressive treatment of cardiac disease and COPD in patients with both conditions have still not been clarified. Given their relevance in terms of prevalence and prognosis, we will focus in this paper on the management of COPD patients with ischemic coronary disease, heart failure and dysrhythmia.

COPD as an endothelial disorder: endothelial injury linking lesions in the lungs and other organs? (2017 Grover Conference Series)

Chronic obstructive pulmonary disease (COPD) is characterized by chronic expiratory airflow obstruction that is not fully reversible. COPD patients develop varying degrees of emphysema, small and large airway disease, and various co-morbidities. It has not been clear whether these co-morbidities share common underlying pathogenic processes with the pulmonary lesions. Early research into the pathogenesis of COPD focused on the contributions of injury to the extracellular matrix and pulmonary epithelial cells. More recently, cigarette smoke-induced endothelial dysfunction/injury have been linked to the pulmonary lesions in COPD (especially emphysema) and systemic co-morbidities including atherosclerosis, pulmonary hypertension, and chronic renal injury. Herein, we review the evidence linking endothelial injury to COPD, and the pathways underlying endothelial injury and the "vascular COPD phenotype" including: (1) direct toxic effects of cigarette smoke on endothelial cells; (2) generation of auto-antibodies directed against endothelial cells; (3) vascular inflammation; (4) increased oxidative stress levels in vessels inducing increases in lipid peroxidation and increased activation of the receptor for advanced glycation end-products (RAGE); (5) reduced activation of the anti-oxidant pathways in endothelial cells; (6) increased endothelial cell release of mediators with vasoconstrictor, pro-inflammatory, and remodeling activities (endothelin-1) and reduced endothelial cell expression of mediators that promote vasodilation and homeostasis of endothelial cells (nitric oxide synthase and prostacyclin); and (7) increased endoplasmic reticular stress and the unfolded protein response in endothelial cells. We also review the literature on studies of drugs that inhibit RAGE signaling in other diseases (angiotensin-converting enzyme inhibitors and angiotensin receptor blockers), or vasodilators developed for idiopathic pulmonary arterial hypertension that have been tested on cell culture systems, animal models of COPD, and/or smokers and COPD patients.

Difference in the craniocaudal gradient of the maximum pixel value change rate between chronic obstructive pulmonary disease patients and normal subjects using sub-mGy dynamic chest radiography with a flat panel detector system

OBJECTIVES

To compare the craniocaudal gradients of the maximum pixel value change rate (MPCR) during tidal breathing between chronic obstructive pulmonary disease (COPD) patients and normal subjects using dynamic chest radiography.

MATERIALS AND METHODS

This prospective study was approved by the institutional review board and all participants provided written informed consent. Forty-three COPD patients (mean age, 71.6±8.7 years) and 47 normal subjects (non-smoker healthy volunteers) (mean age, 54.8±9.8 years) underwent sequential chest radiographs during tidal breathing in a standing position using dynamic chest radiography with a flat panel detector system. We evaluated the craniocaudal gradient of MPCR. The results were analyzed using an unpaired t-test and the Tukey-Kramer method.

RESULTS

The craniocaudal gradients of MPCR in COPD patients were significantly lower than those in normal subjects (right inspiratory phase, 75.5±48.1 vs. 108.9±42.0s^-1cm^-1, P<0.001; right expiratory phase, 66.4±40.6 vs. 89.8±31.6s^-1cm^-1, P=0.003; left inspiratory phase, 75.5±48.2 vs. 108.2±47.2s^-1cm^-1, P=0.002; left expiratory phase, 60.9±38.2 vs. 84.3±29.5s^-1cm^-1, P=0.002). No significant differences in height, weight, or BMI were observed between COPD and normal groups. In the sub-analysis, the gradients in severe COPD patients (global initiative for chronic obstructive lung disease [GOLD] 3 or 4, n=26) were significantly lower than those in mild COPD patients (GOLD 1 or 2, n=17) for both right and left inspiratory/expiratory phases (all P≤0.005).

CONCLUSIONS

A decrease of the craniocaudal gradient of MPCR was observed in COPD patients. The craniocaudal gradient was lower in severe COPD patients than in mild COPD patients.

Phenotyping emphysema and airways disease: Clinical value of quantitative radiological techniques

The pathophysiology of chronic obstructive pulmonary disease (COPD) and Alpha one antitrypsin deficiency is increasingly recognised as complex such that lung function alone is insufficient for early detection, clinical categorisation and dictating management. Quantitative imaging techniques can detect disease earlier and more accurately, and provide an objective tool to help phenotype patients into predominant airways disease or emphysema. Computed tomography provides detailed information relating to structural and anatomical changes seen in COPD, and magnetic resonance imaging/nuclear imaging gives functional and regional information with regards to ventilation and perfusion. It is likely imaging will become part of routine clinical practice, and an understanding of the implications of the data is essential. This review discusses technical and clinical aspects of quantitative imaging in obstructive airways disease.

Pulmonary perfusion quantification with flow-sensitive inversion recovery (FAIR) UTE MRI in small animal imaging

Blood perfusion in lung parenchyma is an important property for assessing lung function. In small animals, its quantitation is limited even with radioactive isotopes or dynamic contrast-enhanced MRI techniques. In this study, the feasibility flow-sensitive alternating inversion recovery (FAIR) for the quantification of blood flow in lung parenchyma in free breathing rats at 7 T has been investigated. In order to obtain sufficient signal from the short T₂ * lung parenchyma, a 2D ultra-short echo time (UTE) Look-Locker read-out has been implemented. Acquisitions were segmented to maintain acquisition time within an acceptable range. A method to perform retrospective respiratory gating (DC-SG) has been applied to investigate the impact of respiratory movement. Reproducibilities within and between sessions were estimated, and the ability of FAIR-UTE to identify the decrease of lung perfusion under hyperoxic conditions was tested. The implemented technique allowed for the visualization of lung parenchyma with excellent SNR and no respiratory artifact even in ungated acquisitions. Lung parenchyma perfusion was obtained as 32.54 ± 2.26 mL/g/min in the left lung, and 34.09 ± 2.75 mL/g/min in the right lung. Application of retrospective gating significantly but minimally changes the perfusion values, implying that respiratory gating may not be necessary with this center-our acquisition method. A decrease of 10% in lung perfusion was found between normoxic and hyperoxic conditions, proving the feasibility of the FAIR-UTE approach to quantify lung perfusion changes.

Developing and Implementing Biomarkers and Novel Imaging in COPD

This article serves as a CME available, enduring material summary of the following COPD9_USA presentations: "Computed Tomography and COPD" Presenter: George R. Washko, MD "CT Imaging in Routine Clinical Practice: Are We Ready for Prime Time?" Presenter: Meilan K. Han, MD "Beyond CT: What MRI can Tell Us about COPD" Presenter: R. Graham Barr, MD.

Probing lung microstructure with hyperpolarized noble gas diffusion MRI: theoretical models and experimental results

The introduction of hyperpolarized gases ((3)He and (129)Xe) has opened the door to applications for which gaseous agents are uniquely suited-lung MRI. One of the pulmonary applications, diffusion MRI, relies on measuring Brownian motion of inhaled hyperpolarized gas atoms diffusing in lung airspaces. In this article we provide an overview of the theoretical ideas behind hyperpolarized gas diffusion MRI and the results obtained over the decade-long research. We describe a simple technique based on measuring gas apparent diffusion coefficient (ADC) and an advanced technique, in vivo lung morphometry, that quantifies lung microstructure both in terms of Weibel parameters (acinar airways radii and alveolar depth) and standard metrics (mean linear intercept, surface-to-volume ratio, and alveolar density) that are widely used by lung researchers but were previously available only from invasive lung biopsy. This technique has the ability to provide unique three-dimensional tomographic information on lung microstructure from a less than 15 s MRI scan with results that are in good agreement with direct histological measurements. These safe and sensitive diffusion measurements improve our understanding of lung structure and functioning in health and disease, providing a platform for monitoring the efficacy of therapeutic interventions in clinical trials.

Improved prediction of lobar perfusion contribution using technetium-99m-labeled macroaggregate of albumin single photon emission computed tomography/computed tomography with attenuation correction

OBJECTIVE

Lung cancer resection can require removal of an entire lobe and, at times, bilobectomy or pneumonectomy. Many patients will also have significantly compromised lung function that requires limiting the extent of surgery or could preclude surgery altogether. The preoperative assessment should include predicted postoperative forced expiratory volume in 1 second (ppoFEV1), because a ppoFEV1 of <40% predicts significantly increased perioperative morbidity. The ppoFEV1 can be estimated by multiplying the preoperative FEV1 by the residual perfused territory percentage, as predicted on planar perfusion scintigraphy (PPS). However, ppoFEV1 using PPS has shown variable correlation with spirometry-measured postoperative FEV1.

METHODS

We propose an improved method for assessing regional lung perfusion in preoperative lung surgery patients. Patients undergo single photon emission computed tomography/computed tomography (SPECT/CT) imaging with attenuation correction using the conventional perfusion agent, technetium-99m-labeled macroaggregate of albumin. The CT image provides information for manual segmentation of each lobe. These segmentations are applied to the SPECT images to determine lobar perfusion. This proposed method was compared with PPS.

RESULTS

This technique was evaluated in 17 patients. As expected, the perfusion contributions of the right and left lungs, calculated from SPECT/CT, correlated closely with those obtained from PPS (Pearson r=0.995). However, the lobar perfusion contributions obtained by PPS and SPECT/CT were significantly different, by 2 methods of comparison (Hotelling's P=1.7×10(-6) and P=1.7×10(-4)).

CONCLUSIONS

This new SPECT/CT technique provides an anatomically more accurate assessment of lobar perfusion. This technique can refine which patients should be operative candidates and allow better prediction of postoperative function in contrast to the anatomically inaccurate planar scintigraphic predictions, which often underestimate the postoperative FEV1. This new technique is expected to have a significant effect on the resectability of patients with lung cancer.

The clinical and economic burden of chronic obstructive pulmonary disease in the USA

Chronic obstructive pulmonary disease (COPD) is the third most common cause of death in the USA. In 2010, the cost of COPD in the USA was projected to be approximately US$50 billion, which includes $20 billion in indirect costs and $30 billion in direct health care expenditures. These costs can be expected to continue to rise with this progressive disease. Costs increase with increasing severity of disease, and hospital stays account for the majority of these costs. Patients are diagnosed with COPD following a multifactorial assessment that includes spirometry, clinical presentation, symptomatology, and risk factors. Smoking cessation interventions are the most influential factor in COPD management. The primary goal of chronic COPD management is stabilization of chronic disease and prevention of acute exacerbations. Bronchodilators are the mainstay of COPD therapy. Patients with few symptoms and low exacerbation risk should be treated with a short-acting bronchodilator as needed for breathlessness. Progression of symptoms, as well as possible decline in forced expiratory volume in the first second of expiration (FEV1), warrant the use of long-acting bronchodilators. For patients with frequent exacerbations with or without consistent symptoms, inhaled corticosteroids should be considered in addition to a long-acting beta2-agonist (LABA) or long-acting muscarinic antagonist (LAMA) and may even consist of "triple therapy" with all three agents with more severe disease. Phosphodiesterase-4 inhibitors may be an option in patients with frequent exacerbations and symptoms of chronic bronchitis. In addition to a variety of novel ultra-LABAs, LAMAs and combination bronchodilator and inhaled corticosteroid (ICS) therapies, other bronchodilators with a variety of mechanisms are also being considered, to expand therapeutic options for the treatment of COPD. With more than 50 new medications in the pipeline for the treatment of COPD, optimal management will continue to evolve and grow more complex as benefits of therapy are balanced with the limitations and needs of each patient.