Precision treatment of COPD based on novel imaging phenotypes: a treatable traits approach

Abstract

Chronic obstructive pulmonary disease (COPD) is a heterogeneous syndrome. Spirometry, while diagnostic, inadequately characterizes disease complexity. This review explores how thoracic imaging, particularly computed tomography and magnetic resonance imaging, may help identify specific pulmonary “treatable traits” that could inform future precision management approaches. We detail four core imaging phenotypes: emphysema, small airway disease, airway mucus plugs, and airway wall thickening. For each, we discuss validated and emerging quantitative biomarkers—such as the Parametric Response Map, total airway count, mucus plug score, Pi10, and PiSlope—that facilitate phenotypic stratification and prognostication. We further describe two distinct disease trajectories (“Tissue-Airway” and “Airway-Tissue”) revealed by progression modeling. Critically, we discuss potential links between these imaging-defined traits to targeted therapeutic strategies, including ultra-fine particle inhalers for small airway disease, mucus clearance strategies, and CT-guided lung volume reduction for emphysema. Despite significant progress, challenges remain in standardizing measurements, validating clinical utility, and integrating imaging biomarkers into routine care. Future integration of artificial intelligence and multimodal imaging holds promise for advancing COPD management towards true personalized medicine.

Introduction

Chronic obstructive pulmonary disease (COPD) has emerged as one of the top three causes of mortality worldwide [1]. It is not a single disease entity but rather a complex, heterogeneous syndrome characterized by significant individual variations in clinical presentation, pulmonary function parameters, and pathophysiological mechanisms [2, 3]. Currently, COPD is defined by the presence of incompletely reversible expiratory airflow limitation, identified via pulmonary function testing, resulting from the combined effects of small airway remodeling and emphysematous destruction of the lung parenchyma [2, 3]. Spirometry forms the cornerstone of COPD diagnosis, with a post-bronchodilator FEV₁/FVC ratio of less than 0.70 (or below the adjusted lower limit of normal) established as the gold standard for confirming persistent airflow limitation. However, with the advancement of COPD assessment methods, the use of spirometry to diagnose COPD has become controversial due to its flawed performance. The main reason is that spirometry alone cannot fully capture all aspects of the heterogeneous nature of COPD. Additionally, studies have shown that some individuals already exhibit significant chronic respiratory symptoms, structural lesions, and/or dysfunction, with poor long-term longitudinal prognosis; however, their lung function, assessed by the FEV₁/FVC ratio, has not yet fallen below 0.70, and thus cannot be diagnosed as COPD. Conversely, a considerable portion of individuals who achieve an FEV₁/FVC < 0.70 do have impaired lung function but lack chronic respiratory symptoms and/or structural lesions. Their rate of lung function decline is slow, and they experience almost no acute exacerbations even without treatment [4, 5].

In recent years, increasing emphasis has been placed on imaging techniques, such as computed tomography (CT) and magnetic resonance imaging (MRI), which non-invasively and precisely reveal a range of structural and functional abnormalities associated with the disease. CT imaging can now provide quantitative data on the airways and lung parenchyma [6]. Similarly, MRI has diverse applications. Hyperpolarized noble gas MRI has been utilized to assess regional ventilation and the degree of alveolar airspace enlargement in COPD patients [7–9]. Despite substantial advancements in imaging research, its integration into routine clinical practice remains limited [10]. Establishing a clear link between imaging findings and clinical outcomes, along with defining its role in longitudinal monitoring, would significantly enhance the clinical value of these techniques.

To address the heterogeneity and complexity of COPD in clinical practice, a management strategy based on identifying “treatable traits” (TTs) has been proposed [11]. These traits can be identified either through phenotypic recognition or via validated biomarkers that provide insights into key pathogenic pathways (endotypes) [12, 13]. Multiple TTs may coexist within a single patient [14], and these traits can change spontaneously over time or in response to therapy. In the field of imaging, such traits specifically refer to therapeutic targets determined based on imaging features and validated imaging biomarkers, and must meet the following three criteria: they are clinically relevant, identifiable and measurable, and treatable with evidence that they can significantly change clinical outcomes. This review will focus on the pulmonary imaging-derived TTs in COPD, with an emphasis on CT and MRI techniques. It will further categorize and stratify these treatable traits and their corresponding treatment responsiveness, aiming to provide novel insights and guidance for clinical management.

Imaging manifestations of COPD

Imaging findings often represent the patient’s underlying pathology [15]. In COPD, imaging can clearly delineate both intrapulmonary and extrapulmonary structures, providing an excellent non-invasive modality for a precise and comprehensive understanding of the disease [10].

Emphysema

Emphysema is a fundamental component in the pathogenesis of COPD, defined as an abnormal, permanent enlargement of airspaces distal to the terminal bronchioles, accompanied by destruction of their walls [16]. On CT, emphysema is quantitatively defined as areas of lung attenuation less than − 950 Hounsfield Units (HU) on inspiratory scans [17, 18]. By visual assessment on CT, emphysema has been classically categorized into three main subtypes: centrilobular, panlobular (also known as panacinar), and paraseptal emphysema [18]. Centrilobular emphysema is the most common form of smoking-related emphysema. It begins in the central portion of the secondary pulmonary lobule, initially appearing as small holes that become more confluent with disease progression. It is typically upper-lobe predominant and characterized by the obliteration or narrowing of distal bronchioles. Panlobular emphysema is classically seen in the lower lobes of patients with alpha-1 antitrypsin deficiency, diffusely involving each secondary lobule, and its severity and age of onset are accelerated in the presence of smoking. Paraseptal emphysema affects secondary lobules adjacent to the mediastinal, costal, and interlobar pleural surfaces, most commonly in the upper lobes [19].

Although visual assessment on CT can determine emphysema subtypes and is an effective tool for estimating the extent of emphysema, it is time-consuming and limited by inter-observer variability. With advances in imaging and the development of unsupervised and deep learning, new classifications of emphysema have been supported and validated by clinical data.

Angelini et al. [20] applied unsupervised machine learning to CT images across two clinical cohorts, using this method to define novel emphysema subtypes. They recategorized emphysema into the following six phenotypes: confluent bronchitis-apical emphysema, diffuse emphysema, senile emphysema, restrictive mixed pulmonary fibrosis and emphysema, obstructive mixed pulmonary fibrosis and emphysema, and vanishing lung syndrome. These novel phenotypes exhibit more nuanced clinical and physiological differences. They not only effectively predict airflow limitation but also show associations with genetics, providing a potential pathway for developing targeted therapies for specific subtypes [20]. The proposal of these new phenotypes better illustrates, from a cross-sectional perspective, the extreme heterogeneity of COPD, highlighting the necessity for future precision medicine.

Furthermore, the Fleischner Society proposed a structured system for the visual classification of parenchymal emphysema [18]. This system uses a six-point ordinal scale to grade parenchymal emphysema as: none, trivial, mild, moderate, confluent, or advanced destructive. Recently, Humphries et al. [21] successfully developed a deep learning system capable of automated, standardized execution of the Fleischner emphysema grading, using the COPDGene cohort. This automated classification not only showed reasonable agreement with visual scores but also demonstrated stronger associations with clinical outcomes such as pulmonary function impairment and mortality risk. This suggests that it may capture imaging features related to prognosis that are overlooked by or difficult to interpret consistently by human observers. This classification assesses the severity of emphysema in COPD longitudinally, placing greater emphasis on clinical validity. Although the aforementioned novel emphysema subtypes exhibit excellent specificity and automation, they lack prospective data validation and remain at a stage of potential clinical promise. In recent years, an increasing number of novel CT-based emphysema subtypes have been proposed. While most claim good clinical relevance, nearly all suffer from inadequate validation. Future research should prioritize validation of their clinical utility, implementation costs, and standardized algorithms, rather than merely generating additional subtypes using novel approaches such as deep learning.

The low proton density of lung tissue, rapid signal decay, and physiological motion from breathing and the heart pose challenges for lung MRI. Despite these inherent difficulties, ultrashort echo time (UTE) MRI is now available for quantifying tissue loss (i.e., emphysema) in COPD patients via an emphysema index [22, 23]. UTE MRI signal intensity measurements have been found to be reproducible and potentially as useful as quantitative CT [24, 25]. Moreover, T₂* mapping using UTE in COPD correlates well with spirometric values (FEV₁ and FEV₁/FVC) and the transfer factor (Tico) [26, 27]. Therefore, UTE-MRI can serve as a biomarker for the extent and severity of COPD (and emphysema). Furthermore, hyperpolarized (HP) gas diffusion MRI provides images of acinar microstructure sensitive to emphysema, including early emphysematous changes in asymptomatic smokers [27], and is highly reproducible [28, 29].

Thus, advances in both CT and MRI not only provide more precise biomarkers for quantifying emphysema but also offer potential for the precise treatment of emphysema in COPD.

Small airway disease

Small airway disease is an early and central event in COPD. Studies have confirmed that significant narrowing and loss of small airways (diameter < 2 mm) occur during the course of the disease [30]. Often colloquially referred to as the “silent zone” of the lungs, it is estimated that approximately 75% of small airways must be damaged before changes become detectable by standard pulmonary function tests [31]. It has been reported that compared to controls, patients with mild COPD have a reduced number of small airways, and this number progressively decreases with advancing GOLD stages [32]. Therefore, accurate clinical identification of small airway disease is crucial for the early treatment of COPD patients.

However, conventional high-resolution CT (HRCT) is unable to resolve these small airways [33]. In imaging, small airway narrowing is defined as a reduction in the number of small airways with diameters between 2.0 mm and 2.5 mm, as detected by micro-CT compared to control subjects [32]. However, as micro-CT is not routinely used in clinical practice, this poses a significant challenge for observing small airway disease with standard clinical CT.

A breakthrough in recent years has successfully addressed this challenge. Galbán et al. [34] proposed and validated a novel CT-based imaging biomarker called the Parametric Response Map (PRM) for diagnosing COPD. PRM is a voxel-wise image analysis technique that enables the independent, quantitative, and spatially localized visualization of two core pathological components of COPD—functional small airway disease (fSAD) and emphysema—through paired analysis of registered inspiratory and expiratory chest CT scans. PRM can clearly distinguish patients with fSAD-dominant, emphysema-dominant, or mixed-type COPD, even when their conventional pulmonary function measures (such as FEV₁) are similar [34]. They defined fSAD as areas of low attenuation on expiration measuring < −856 HU but on inspiration measuring > −950 HU. This makes the imaging-based identification of fSAD clinically feasible. This finding was substantiated by Vasilescu et al. [35], who used direct evidence from micro-CT to show that the chest CT biomarker fSAD^PRM correlates with terminal bronchiolar (TB) loss, as well as narrowing, thickening, and obstruction of remaining TBs in COPD—even in regions without emphysema. This provides more direct evidence for the role of PRM while further characterizing the features of fSAD. Therefore, they propose renaming the PRM classification from “functional small airway disease” to “small airway disease (SAD)” to more directly reflect its association with anatomical pathology. However, quantifying fSAD requires an additional expiratory phase CT scan, which limits its clinical applicability due to increased cost, radiation exposure, and technical complexity.

To address this, Chaudhary et al. [36] employed a novel generative artificial intelligence method to assess fSAD from a single inspiratory chest CT scan acquired at total lung capacity (TLC). They found that the derived single-volume metric, fSAD^TLC, correlated strongly with the traditional gold-standard metric fSAD^PRM (which requires both inspiratory and expiratory scans) and demonstrated clear clinical significance (e.g., association with worse lung function and higher mortality). Furthermore, relying solely on a single inspiratory CT, fSAD^TLC exhibited superior reproducibility. This holds potential for facilitating the broader clinical adoption of PRM-based assessments. It should be noted that this study involved only short-term follow-up; long-term follow-up is warranted, and more large-scale prospective studies are required to further demonstrate its clinical utility. Overall, despite the great potential of PRM in clinical research, it still faces many limitations in practical application. These include increased radiation dose, prolonged scan time, high demands on patient cooperation, variability in data processing, and the complexity of associations with physiological parameters. All these limitations hinder the widespread adoption and precise application of PRM [37, 38]. Future efforts should integrate topological analysis methods into CT-PRM technology, which may enhance diagnostic performance while reducing radiation dose [39].

In recent years, proton MRI has emerged as a promising, radiation-free alternative to CT for characterizing and monitoring COPD [40]. Airway morphological assessment using MRI is feasible [41] and has been shown to be reliable and reproducible in COPD patients [41]. SAD (both in COPD and other respiratory diseases) can also be sensitively depicted by MRI [42, 43]. Furthermore, SAD can be directly assessed by quantifying ventilation defects on MRI [10].

Airway mucus obstruction

Historically, excessive mucus secretion in COPD was often considered a relatively benign process, with its role in airflow obstruction and disease progression thought to be minimal [44, 45]. However, findings from multiple recent epidemiological studies demonstrate a significant association between mucus hypersecretion and adverse patient outcomes in COPD [46, 47], indicating that research on airway mucus may hold important therapeutic implications.

Under normal circumstances, CT can only evaluate obstructive mucus plugs in large to medium-sized airways (2–10 mm) [33]. Mucus plugs are typically identified as dense opacities within the airway lumen, connected to a patent airway on consecutive axial CT slices. These opacities exhibit lower radiodensity than adjacent blood vessels, and a mucus plug is defined as causing complete bronchial occlusion [48]. Diaz et al. [49] evaluated airway mucus plugs in 4363 patients. They found that a substantial proportion (40.7%) of these patients had at least one segmental obstructive mucus plug on baseline CT (21.8% had 1–2 segments, 18.9% had ≥ 3 segments), demonstrating that mucus plugs are highly prevalent in COPD patients. Furthermore, after adjusting for a history of chronic bronchitis, current asthma, coronary artery disease, frequency of acute exacerbations, and even the validated BODE mortality risk index, the presence of mucus plugs remained an independent risk factor for mortality. Concurrently, a prospective study by Li et al. [50] showed that the presence of airway mucus plugs was significantly associated with acute exacerbations of COPD, and the incidence of exacerbations increased markedly with higher mucus plug scores. Additionally, they confirmed that patients with a mucus plug score ≥ 4 had lower BMI, worse lung function, more severe symptom scores (CAT, mMRC), and a greater history of prior severe exacerbations. A study published in the New England Journal of Medicine not only further confirmed the association between mucus plugs and lung function decline, but also demonstrated that the dynamic changes in mucus plugs determine the disease trajectory. Specifically, patients whose mucus plugs resolved on follow-up imaging experienced a significantly slower rate of lung function decline, whereas patients who developed new mucus plugs showed a markedly accelerated decline. Furthermore, smoking was found to substantially amplify the contribution of mucus plugs to airflow obstruction [51].

Moreover, the impact of airway mucus plugs on these adverse outcomes can be independent of the presence of emphysema or the reduction in small airways. Dunican et al. demonstrated [46] that both mucus plug score and percentage of emphysema were independently associated with lower FEV₁, lower FVC, and poorer oxygenation. Analysis of follow-up CT scans one year later in 100 COPD patients indicated that the mucus plug phenotype was highly stable. Tanabe et al. [52] conducted a large-scale, multicenter prospective study using two independent Japanese COPD cohorts. They found that the total airway count (TAC) was significantly lower in the high mucus plug group compared to the no mucus plug group, indicating an association between mucus plugs and reduced airway branching. The mucus plug score showed no correlation with emphysema severity but was associated with elevated parameters reflecting small airway dysfunction (SAD%), suggesting that mucus plugs may cause or be accompanied by air trapping. They further distinguished between reduced airway branching and mucus plugs, finding that mucus plugs were primarily associated with increased mortality, while reduced airway branching was mainly linked to declines in lung function, worsening symptoms, and loss of health independence.

Mucus obstruction also occurs in small airways [53, 54]. Research by Hogg et al. [47] found that the lumens of small airways contained inflammatory exudates with mucus, and this was an independent risk factor for predicting mortality following lung volume reduction surgery in patients with advanced COPD. They also confirmed that corticosteroid therapy had no significant effect on reducing small airway wall thickening or clearing the inflammatory mucus plugs within the lumen. However, the precise pathological structure of small airway obstruction was initially unclear [55]. In the latest research [56], micro-CT was used to provide a detailed description of the types of small airway obstruction in smokers with COPD. Three primary obstruction types were identified at the terminal bronchiolar level: webs structures, occlusions, and collapses (Labyrinthine structures were described as very thin, fragile, net-like structures traversing but not completely blocking the airway lumen. Occlusions were identified as thicker, denser structures crossing the lumen and interrupting its patency while preserving the airway lining. Collapse was characterized by airway narrowing with a segment of the lumen disappearing). These obstructions were predominantly composed of mucus plugs infiltrated with immune cells, particularly histiocytes and neutrophils, which differs markedly from the eosinophil-dominated mucus plugs seen in asthma [57]. Notably, histological examination of these mucus plugs revealed no evidence of epithelial remodeling, except for the presence of goblet cell hyperplasia in one COPD sample. The absence of goblet cell hyperplasia may be attributed to the relative scarcity of goblet cells and the predominance of Clara cells in the terminal bronchioles [31]. Consequently, it is likely that the mucus originates from larger airways where mucociliary clearance is impaired, leading to its accumulation as airway plugs [58, 59]. However, the sample size in this study may have been limited. While micro-CT provides unparalleled detail about the pathological composition of small airway obstruction, its ex vivo nature and limited availability render it unsuitable for routine clinical use. For clinical practice, the conventional CT mucus plug score—though limited to larger airways—remains the most pragmatic imaging biomarker for identifying patients with a ‘mucus plugging’ treatable trait, particularly given its validated associations with mortality, exacerbations, and lung function decline. The micro-CT findings should be viewed as providing the pathological rationale for targeting mucus, rather than as a required clinical tool. To demonstrate that small airway obstruction indeed originates from airway mucus plugs, further clinical studies with larger sample sizes are needed to identify the true contributors to small airway obstruction.

In clinical practice, physicians have traditionally focused more on “symptomatic mucus”—i.e., COPD patients presenting with cough and sputum production [45, 60]. However, recent research may challenge this limited perspective. A recent study by Mettler et al. [61] provides a detailed comparison between “silent mucus” (CT-detectable mucus obstruction in patients without mucus-related symptoms) and symptomatic mucus. They found that among COPD patients without symptoms of cough or sputum, a significant proportion (up to 36%) had CT-detectable silent mucus plugs, indicating that mucus-related symptoms do not fully align with the underlying pathology of mucus obstruction. Independent risk factors for silent mucus plugs, compared to symptomatic mucus obstruction, were older age, female sex, and Black race. Despite the absence of clinical symptoms, silent mucus plugs (especially with a score ≥ 3) were also significantly associated with worse disease metrics compared to COPD patients without mucus obstruction [61]. Furthermore, the demographic and structural characteristics of patients with silent mucus plugs (e.g., older age, female sex, more prominent emphysema) differed markedly from those of traditional chronic bronchitis patients (typically younger males with heavier smoking history) [3], suggesting that airway mucus plugs may represent a distinct COPD phenotype encompassing features of both chronic bronchitis and emphysema, rather than being merely an imaging manifestation of chronic bronchitis alone [61].(Table 1; Fig. 1).

Table 1.

Comparative characteristics of “silent” vs. “symptomatic” airway mucus plugs in COPD

Comparative Dimension	“Silent” Mucus Plugs	“Symptomatic” Mucus Plugs	Key Reference (s)
Demographic Features	Independent risk factors include older age, female sex, and Black race. The patient profile differs from that of traditional chronic bronchitis	Typically associated with heavy smoking history and male sex (the classic chronic bronchitis phenotype)	[61]
CT Mucus Plug Score	A score ≥ 3 is significantly associated with worse disease metrics (e.g., lung function). The score does not directly correlate with clinical symptoms	The mucus plug score correlates with the patient’s sputum-producing symptoms; a higher score usually indicates more pronounced respiratory symptoms	[49, 50, 61]
Presence of Symptoms	Absence of typical chronic bronchitis symptoms (chronic cough and sputum). Dissociation between imaging findings and clinical symptoms is its defining feature	Presence. Consistent with the clinical definition of chronic bronchitis, characterized by persistent cough and sputum production	[45, 61]
Association with Lung Function Decline	An independent risk factor for accelerated FEV₁ decline. Longitudinal studies show that newly developed plugs are associated with a faster rate of decline, while resolution of plugs is linked to a slower decline. Smoking amplifies this effect	Associated with increased risk of exacerbations and worse symptom scores (CAT, mMRC), which indirectly affect lung function trajectory and patient health status	[50, 51]
Mortality Risk	An independent risk factor for all-cause mortality. This association persists even after adjusting for history of chronic bronchitis, exacerbation frequency, and the BODE index	Typically serves as a marker of disease severity and chronic inflammatory activity. Its mortality risk is often mediated indirectly through increased exacerbations and respiratory failure	[49]
Therapeutic Implications	1. Identification: Easily missed by routine clinical assessment. Requires active screening via CT imaging, especially in high-risk groups (e.g., specific imaging phenotypes, older females) 2. Treatment: Highlights the critical importance of smoking cessation (which can significantly attenuate its harm). As the pathophysiology may involve different mechanisms of mucociliary dysfunction, response to conventional mucoactive agents requires specific study.	1. Identification: Easily identifiable by clinical symptoms, representing a clear target for management. 2. Treatment: The primary target population for airway clearance techniques and mucoactive medications (e.g., hypertonic saline, mucoregulators). However, note that recent high-quality RCTs (e.g., the CLEAR study) indicate limited overall efficacy of pharmacologic agents alone in preventing exacerbations	[51, 55, 61, 62]

Fig. 1.

Symptomatic vs. silent mucus plugs in COPD. Mucus plug pathology can be dissociated from cough/phlegm symptoms. While the left panel shows symptomatic cases, the right panel indicates that up to 36% of asymptomatic COPD patients have CT-detectable mucus plugs, which are associated with worse lung function and higher exacerbation risk-even in the absence of cough/phlegm

MRI also has applications in imaging airway mucus [63, 64], although research has been more extensive in asthma [64].

It is important to recognize that mucus hypersecretion, chronic bronchitis, CT-visible mucus plugs, and the mucus burden in bronchiectasis represent interrelated but not interchangeable pathological states. Mucus hypersecretion is a pathophysiological process characterized by excessive production of airway mucus [65]. Chronic bronchitis, by contrast, is a clinical diagnosis defined by chronic cough with sputum production for at least three months per year over two consecutive years, which may occur with or without radiologically detectable mucus plugs [66]. CT-visible mucus plugs are imaging findings of complete bronchial occlusion by inspissated secretions, which can be identified and scored even in patients who do not meet the clinical criteria for chronic bronchitis [67]. In COPD with coexistent bronchiectasis, the mucus burden is further complicated by impaired mucociliary clearance due to permanent airway dilatation, leading to a distinct secretory phenotype that may require different management strategies [68]. These distinctions are clinically meaningful, as the identification of imaging-defined mucus plugging alone-independent of chronic bronchitis symptoms-is increasingly recognized as a treatable trait with prognostic significance [69].

Airway wall thickening

Airway wall thickening contributes equally to airflow obstruction as emphysema in COPD patients and is often considered an independent COPD phenotype [70]. In COPD, airway wall thickening results from goblet cell hyperplasia, submucosal thickening, as well as adventitial inflammation and fibrosis, clinically manifesting as chronic bronchitis [71].

However, precise quantification of airway thickening is relatively difficult [72], as airway changes are often localized and heterogeneous [71]. Initial assessments used segmental or subsegmental airway wall thickness, but these are limited by the significant influence of individual height and body size on airway dimensions. This limitation can be partially addressed by calculating the wall area percentage [73]. Furthermore, these metrics rely on the assumption that a single measurement from a more central location reflects distal airway remodeling, which further restricts their applicability. To overcome this, Pi10 was developed [74, 75]. Pi10 is calculated by plotting the square root of the wall area against the corresponding inner perimeter for all visible airways and then determining the wall thickness of a hypothetical airway with an inner perimeter of 10 mm [74]. Pi10 is a summary metric for all visible branches in the airway tree, with higher values indicating thicker airway walls.

Higher Pi10 is associated with multiple cross-sectional clinical outcomes, including lower lung function [76–78], more severe dyspnea [79], a higher prevalence of chronic bronchitis [80], lower functional capacity [81], and worse respiratory-related quality of life [81, 82]. Higher Pi10 also predicts increased exacerbation frequency [83], a faster rate of FEV₁ [76, 78, 83], and higher all-cause mortality [78, 83]. However, Pi10 assessment is based on the assumption that the relationship between wall thickness and lumen perimeter follows a fixed slope, which is similar across individuals. Given that this relationship varies between individuals, the slope (PiSlope) was investigated more precisely by Bhatt et al. using the COPDGene cohort. They found that a lower PiSlope was associated not only with more severe dyspnea (mMRC score), worse quality of life (SGRQ score), and lower functional capacity (6-minute walk distance), but also predicted higher exacerbation frequency, a faster FEV₁ decline, and higher all-cause mortality. These associations remained significant for most outcomes even after adjusting for the traditional metric Pi10, indicating that PiSlope provides incremental prognostic information beyond Pi10 [84]. Furthermore, PiSlope may reflect the heterogeneity of airway remodeling distribution along the airway tree (e.g., disproportionately thickened distal airways), which might align more closely with the pathophysiology of COPD (e.g., early distal disease) [84]. However, although the definitions of Pi10 and PiSlope are clear, considerable methodological heterogeneity exists across studies in their actual calculation, which constitutes a core limitation of their application. Studies have shown that at least ten literature-based computational methods have been used for Pi10 and PiSlope, with significant differences in key parameters. For example, some methods measure only airways of specific sizes (e.g., Patel and Van Tho methods), others restrict measurements to specific airway generations (e.g., Gietema and Park methods), while still others cover a broader range of airway sizes and generations (e.g., Bhatt, Jobst, and Smith methods). This heterogeneity in calculation methods directly leads to “substantial variability” in Pi10 and PiSlope values across studies, making cross-study comparisons and result interpretation extremely difficult, and limiting their potential as standardized biomarkers [85].

For MRI, COPD-related pathology within lobar and segmental bronchi, such as bronchial wall thickening, mucus plugs, and/or bronchiectasis, can be observed on T2-weighted images [40], however, supporting research remains limited.

Although airway wall remodeling can occur in some COPD patients without chronic bronchitis symptoms [71], clinical quantification of airway thickening without added radiation could improve patient stratification, prognosis prediction, and identification of high-risk patients.

Heterogeneous disease trajectories in COPD

A key aspect of COPD heterogeneity is that patients often present with multiple clinical phenotypes, which may develop in a specific temporal sequence [86]. Building on the summarized imaging phenotypes and recent research, we can now delineate more detailed disease trajectories to better identify treatable pulmonary traits in COPD.

A recent breakthrough study applied a novel machine learning tool, “Subtype and Stage Inference” (SuStaIn), to identify patient subgroups with distinct longitudinal progression patterns. The study concluded that COPD development follows two primary trajectories: a “Tissue-Airway” pattern, where small airway abnormalities and emphysema precede large airway wall abnormalities; and an “Airway-Tissue” pattern, characterized by large airway wall abnormalities occurring prior to emphysema and small airway dysfunction. Approximately 70% of patients exhibited the Tissue-Airway pattern, demonstrating worse lung function with a nonlinear decline across SuStaIn stages—more rapid in early stages. The remaining 30% displayed the Airway-Tissue pattern, showing relatively preserved lung function that declined linearly with disease progression [87]. It should be noted that the SuStaIn-derived trajectories are computationally inferred from cross-sectional data and have not yet been validated in prospective longitudinal cohorts with serial imaging. The mechanistic sequence described below represents a synthesis of pathological and imaging findings that, while biologically plausible, remains an active area of investigation.

Based on this classification, a tentative model of disease progression may be proposed, though it should be interpreted as a working hypothesis derived from cross-sectional studies rather than an established longitudinal sequence. In the Tissue-Airway subtype, as a result of exposure to toxic particles small airways are thought to initially develop webs obstruction, where mucus and inflammatory exudates traverse but do not fully occlude the lumen. It has been proposed that this may evolve into occlusions and collapse lesions. The occlusions pattern features denser materials, such as mucus plugs with immune cell infiltration, that interrupt luminal patency. The collapse pattern is characterized by airway narrowing due to loss of alveolar attachments. Mucous obstruction and associated inflammation may ultimately lead to terminal bronchiolar obliteration and collapse, inducing adjacent emphysematous destruction [53, 56, 88]. This may explain the greater disease activity in early SuStaIn stages. Subsequently, factors like airflow obstruction may drive increased mucus secretion and heightened inflammatory responses in larger airways [56, 89–91]. Conversely, for the Airway-Tissue subtype, one possible scenario is that large airway wall thickening or mucus plugs develop first [48, 61].Under this model, excess mucus could extend into small airways, leading to their obliteration or collapse, and thereby triggering inflammation and subsequent tissue destruction/emphysema. This manifests as large airway abnormalities preceding significant lung function impairment. Notably, mucus plugs in small airways are often infiltrated by neutrophils [56] (Fig. 2).

Fig. 2.

Exemplary diagram of the two COPD disease trajectories: airway-tissue pattern and tissue-airway pattern

Imaging-defined treatable traits in COPD

Before discussing individual imaging-defined traits, it is important to recognize that the clinical significance and management of any given trait depend, in part, on overall COPD severity. The Global Initiative for Chronic Obstructive Lung Disease (GOLD) classification remains the standard for staging disease severity and guiding initial pharmacotherapy [92]. In general, GOLD stage 1–2 patients tend to show a higher proportion of small airway disease and milder emphysema, whereas GOLD stage 3–4 patients have a progressively greater emphysema burden, more extensive mucus plugging, and a higher prevalence of comorbid pulmonary hypertension or bronchiectasis [92]. Consequently, the same imaging trait may call for different therapeutic approaches at different stages: for example, an emphysema-dominant pattern in a patient with GOLD stage 1 may be managed with bronchodilators and risk-factor modification alone, while in GOLD stage 3–4 the same pattern, if accompanied by severe hyperinflation, may warrant evaluation for lung volume reduction [92]. The imaging-based treatable-traits framework described below is therefore intended to be integrated with, rather than replace, standard GOLD-based assessment. Throughout this section, the reader should consider whether the treatment strategies discussed apply primarily to mild–moderate disease, to severe–very severe disease, or across the full spectrum.

It is important to acknowledge that the translation of imaging-defined treatable traits into therapeutic strategies operates along a spectrum of evidence. While the relationship between emphysema heterogeneity and lung volume reduction is well-established, the associations between specific imaging patterns (e.g., PiSlope or fSAD) and newly emerging therapies (e.g., specific biologics or airway clearance modalities) remain at a plausible or exploratory stage. Therefore, clinical decision-making should integrate imaging phenotypes with conventional biomarkers such as blood eosinophil counts and the frequency of exacerbations. Numerous chest CT features are emerging as biomarkers for COPD. Quantitative CT (QCT) can reliably characterize the COPD [15]. Based on the pulmonary pathologies described above, detailed COPD TTs can be identified and may enable their targeted treatment based on well-correlated imaging biomarkers [93]. Furthermore, we have classified the relevant imaging biomarkers into three levels: Level 1 includes established imaging biomarkers that can serve as identifiable TTs to directly guide clinical treatment; Level 2 includes promising TTs that have the potential to guide treatment or have already been used to assess treatment response, but lack sufficient evidence to demonstrate their role in clinical decision-making; and Level 3 includes potential imaging biomarkers that have only been validated in cross-sectional studies or are newly developed (Table 2).

Table 2.

Imaging-defined treatable traits in COPD: a framework of phenotypes, biomarkers, and therapeutic strategies

Imaging-Defined Phenotype (Treatable Trait)	Key Imaging Biomarkers	Evidence Level Classification for Imaging Biomarkers	Current Evidence/Consensus-Based Treatment Strategies	Controversial or Limited-Evidence Strategies	Future Research Directions and Potential Therapies
Emphysema-Dominant	CT: Emphysema index (LAA%950) [17, 18, 94], Percentile index (Perc15) [95], PRMemph [34], SWES [96]. MRI: Apparent Diffusion Coefficient (ADC) and Mean Linear Intercept (Lm) from hyperpolarized gas diffusion MRI [27, 97–99].	Established TTs: Emphysema index, Percentile index Promising TTs: PRMemph, SWES Potential biomarkers: ADC, Lm	Foundation: Long-acting bronchodilators (LAMA/LABA), pulmonary rehabilitation, long-term oxygen therapy (if hypoxic) [100]. Intervention (for severe heterogeneous disease): Lung volume reduction surgery or endobronchial valve placement. [101–103].	ICS offer limited benefit in this phenotype and may increase pneumonia risk [104]. Conventional mucoactive agents are not a primary treatment direction.	Anti-elastase therapies, regenerative therapies targeting alveolar repair. MRI VDP-guided precision intervention [105]
Small Airway Disease-Dominant	CT: PRMfSAD [34, 35], TAC [32, 42], AI-based single-inspiratory assessment (fSADTLC) [36]. Direct Imaging: EB-OCT for measuring airway lumen area [106]	Established TTs: NA Promising TTs: PRMfSAD, TAC, EB-OCT Potential biomarkers: fSADTLC	Pharmacotherapy: Dual bronchodilators (LAMA/LABA), Fixed-dose triple therapy for COPD (ICS/LAMA/LABA) [107]	Conventional inhaler formulations have limited deposition in small airways. Systemic corticosteroids have been shown to be ineffective in reducing small airway wall thickening or clearing inflammatory mucus plugs [47]. Lack of high-quality evidence for oral drugs improving small airway function in COPD [108]	Phenotype-specific clinical trial design based on PRM classification. Targeted therapies for neutrophil-infiltrated mucus plugs in small airways (e.g., NETs inhibitors) [56]
Airway Mucus Plugs	CT: Mucus plug score (number of involved bronchial segments) [48, 49]	Established TTs: NA Promising TTs: Mucus plug score Potential biomarkers: NA	Cornerstone: Smoking cessation significantly attenuates the contribution of mucus plugs to airflow obstruction [51].	Pharmacotherapy: Mucoactive agents (e.g., hypertonic saline, mucoregulators like tiotropium) [109–116], but note their limited overall evidence for reducing exacerbations [62].	Novel agents aimed at restoring mucociliary clearance (e.g., ENaC inhibitors) [58, 59], biologics targeting specific mucus components (e.g., MUC5AC). Targeted lung denervation [117]. Bronchial rheoplasty [118]. Metered cryospray [119]
Airway Wall Thickening	CT: WA% [73, 120], Pi10 [74, 75], PiSlope [84]	Established TTs: NA Promising TTs: WA%, Pi10 Potential biomarkers: PiSlope	Core Pharmacotherapy: Combination regimens containing inhaled corticosteroids (e.g., triple inhalers). Preliminary OCT studies suggest such treatment may increase small airway lumen area and decrease wall area percentage, indicating potential reversal of remodeling [121].	NA	Validation of the prognostic value of PiSlope for predicting disease progression and treatment response [86, 122]. Exploration of anti-fibrotic drugs (e.g., nintedanib) in this phenotype; investigation of biologics targeting specific endotypes (e.g., Tezepelumab for type 2 inflammation, which has shown efficacy in improving small airway remodeling in asthma) in COPD [123].

Imaging biomarkers and therapeutics for emphysema

The most common method for quantifying emphysema is threshold-based low-attenuation area analysis. The most frequently used quantitative metric is the percentage of low-attenuation areas (LAA%), specifically the emphysema index (LAA%_950insp) derived by applying a threshold of −950 HU to inspiratory CT scans. This index, expressed as the relative volume of affected lung parenchyma, has been pathologically validated as a reliable measure of emphysema [94]. Another common densitometry-based method is the percentile index, most notably the 15th percentile (Perc15) [95]. Both metrics have been extensively validated for accurately reflecting air trapping caused by parenchymal destruction and correlate with lung function and other relevant parameters [124–126].

However, density thresholds alone cannot fully capture the morphological features of emphysema. Therefore, morphology-based metrics have been developed, such as the bullous index (BI), which distinguishes bullous from diffuse emphysema [127]. More importantly, paired inspiratory and expiratory CT analyses provide functional information. Quantitative assessment of air trapping (e.g., mean lung density, kurtosis on expiratory CT) and PRM techniques can separately quantify emphysema. Their ability to assess airflow limitation may surpass that of inspiratory LAA% alone [34].

With advances in artificial intelligence, more imaging biomarkers based on deep learning are being developed. For example, the Slice-wise Whole-lung Emphysema Score (SWES), derived via neural networks, significantly outperforms traditional threshold-based methods (e.g., LAV₉₅₀) and radiologist visual scores in identifying visual emphysema and predicting airflow obstruction, thereby avoiding the limitations of threshold methods [96].

Furthermore, MRI offers insights into systemic effects. Hyperpolarized gas (³He or ¹²⁹Xe) diffusion MRI provides unique information on alveolar microstructure, such as the Apparent Diffusion Coefficient (ADC) and Mean Linear Intercept (Lm). These metrics are more sensitive than CT densitometry for detecting early emphysema, particularly in alpha-1 antitrypsin deficiency [97–99]. The ventilation defect percentage (VDP) from ³He-MRI is a powerful tool for assessing regional ventilation [105]. Overall, apart from LAA%950insp, which has been clearly validated as an imaging biomarker to guide treatment, other imaging biomarkers may only be at the level of correlation with clinical indicators or prognostic prediction [128]. In the future, a large number of large-scale prospective studies are still needed to validate their role in clinical treatment decision-making.

Pharmacotherapy for emphysema includes bronchodilators, corticosteroids, and long-term oxygen therapy. Lung volume reduction (LVR) surgery or endobronchial valve placement plays a significant role in treatment [101, 102]. In this context, CT is crucial for identifying the anatomical distribution of emphysema, the presence of collateral ventilation, and fissure integrity. These parameters guide the selection of optimal target regions for LVR or valve placement and help predict postoperative outcomes [103, 129].

Imaging biomarkers and therapeutics for small airway disease

Small airway disease is relatively difficult to identify and is typically assessed indirectly on CT. A well-established biomarker is the TAC, defined as the sum of visible airways from the 3rd to the 8th generation [32, 130, 131]. HRCT can also evaluate direct or indirect signs of small airway pathology. Direct signs primarily include centrilobular nodules and tree-in-bud opacities, while indirect signs encompass mosaic attenuation, wedge-shaped ground-glass opacities, and cylindrical bronchiectasis [132].

In contrast, Optical Coherence Tomography (OCT) is a three-dimensional tomographic technique capable of reaching the 7th to 9th bronchial generations, enabling definitive identification and quantification of airway wall layers [106]. A 2018 study utilizing Endobronchial OCT (EB-OCT) to analyze small airway metrics in COPD patients found significant positive correlations between OCT-derived measurements and the percentage of small airway area, as well as with resonance frequency and respiratory resistance (5–20 Hz) measured by forced oscillation technique [106].

Although imaging indicators have the advantage of directly visualizing small airway disease, current imaging techniques may increase costs, radiation exposure, and even invasiveness [35, 133]. In clinical practice, the treatment of small airway disease should be decided by integrating other diagnostic tests, such as lung function parameters for small airway dysfunction (FEF25%–75% ≤80% predicted or below the lower limit of normal [LLN], FEV1/SVC ≤ 0.70, FEV3/FEV6 ≤ LLN, FEV3/FVC ≤ LLN) or impulse oscillometry (IOS) parameters (R5–R20, area of reactance [AX], and Fres), to guide comprehensive treatment decisions for SAD [134–137].

Regarding treatment, a study by Tse et al. demonstrated that oral high-dose N-acetylcysteine (NAC) administered for 16 and 52 weeks significantly improved small airway function in COPD patients compared to placebo, showing marked enhancements in both reactivity and airway resistance [108]. Furthermore, research by Xiong et al. in 2018 indicated that COPD patients treated with a combination of tiotropium and theophylline showed significant improvement in small airway function at 3 and 6 months. At 6 months, the combination therapy group exhibited significantly better small airway function than the tiotropium monotherapy group [138]. Nevertheless, there remains a lack of high-quality evidence to demonstrate that oral medications improve small airway function in patients with COPD.

The development of ultra-fine particle formulations is particularly important. A study by Dave et al., using Functional Respiratory Imaging (FRI), found that the GF pMDI (budesonide/glycopyrronium/formoterol fumarate) demonstrated the highest total lung deposition (> 54%) and highest total drug delivered at both tested flow rates (30 and 60 L/min). The BDP/G/F pMDI (beclomethasone dipropionate/glycopyrronium/formoterol fumarate) showed moderate total lung deposition (~ 40%), but its deposition was skewed toward small airways (lowest C: P ratio), also indicating co-deposition. In contrast, the FF/UMEC/VI DPI (fluticasone furoate/umeclidinium/vilanterol) had the lowest total lung deposition (especially at 30 L/min), with deposition predominantly in the large airways. The deposition of its three active ingredients was inconsistent, with the ICS component depositing the least, particularly in small airways, potentially due to its larger particle size and lower fine particle fraction (FPF) [139].

Emerging research suggests that small airway obstruction is often composed of neutrophil-infiltrated mucus plugs [56]. Therefore, when imaging or pulmonary function tests indicate small airway disease, the consideration of anti-mucous agents may be warranted.

Imaging biomarkers and treatment for airway wall thickening and airway mucus plugs

There are relatively few biomarkers specific to airway mucus in COPD. Mucus plugs are often only visible on CT in larger airways [140]. The mucus plug score is currently a common biomarker, demonstrating significant clinical value through its association with mortality and lung function decline [48, 49]. However, the airway mucus score can currently only serve as a prognostic indicator. Only a small number of asthma-related studies have shown that biologics (such as dupilumab and mepolizumab) can improve the airway mucus score in patients [141]; however, no studies to date have confirmed that the airway mucus score can be used as an indicator for clinical treatment. In the future, larger prospective studies are still needed to validate its value in guiding treatment.

For quantifying airway wall thickening in COPD, Wall Area Percentage (WA%) and Pi10 are two core imaging biomarkers [75]. Although both aim to reflect thickening, their calculation methods and physiological implications differ fundamentally. WA% directly measures the ratio of the wall area to the total cross-sectional area of a specific airway, typically a segmental bronchus [120].

Both WA% and Pi10 correlate with lung function decline, increased symptoms, and poor prognosis, but the strength and pattern of these associations vary [77, 142]. Multiple studies confirm that Pi10 generally shows a stronger and more consistent inverse correlation with lung function parameters (e.g., FEV₁, FEV₁/FVC), especially when calculated from a larger number of measured airways [76, 77]. WA% is also associated with airflow limitation, chronic bronchitis symptoms [143], and health-related quality of life (e.g., SGRQ score) [120], suggesting WA% and Pi10 contribute to clinical phenotypes from different dimensions.

PiSlope, a newer biomarker, may better reflect abnormalities across the airway tree compared to Pi10 [122], but related research remains limited. Notably, both Pi10 and PiSlope are primarily prognostic biomarkers, used to predict FEV₁ decline and mortality risk. Their utility as predictive biomarkers for selecting specific therapies (e.g., which patient benefits from biologic or anti-remodeling treatments) remains largely unproven and requires prospective validation [144].

Burgel et al. argue that mucus hypersecretion should be a treatment target for all COPD patients, including those with chronic cough but no sputum production [55]. Currently, smoking cessation represents the most effective intervention for airway mucus [51]. Apart from smoking cessation, other clinical treatment decisions need to be based on a comprehensive assessment of imaging findings, the presence of cough and sputum, pulmonary function tests, and inflammatory endotypes [92].

Pharmacological mucoactive drugs are categorized by mechanism:

1. Expectorants/Hydrators: e.g., hypertonic saline (3%−7%) and mannitol, which hydrate airway surfaces via osmosis, effectively thinning mucus and enhancing clearance, with demonstrated clinical benefits in CF and COPD [109–111].

2. Mucolytics: e.g., N-acetylcysteine (disrupts disulfide bonds) and Dornase alfa (hydrolyzes DNA), which reduce mucin viscoelasticity by degrading the polymer network [112, 113].

3. Mucoregulators: e.g., anticholinergics (tiotropium) and long-acting β₂-agonists, which act by reducing mucus secretion or improving ciliary function [114, 115]. Tiotropium has shown symptomatic improvement in mucus hypersecretion in conditions like diffuse panbronchiolitis unresponsive to macrolides [116].

In the recent CLEAR study, Patients were randomized 1:1:1:1 to four groups: hypertonic saline, hypertonic saline combined with carbocisteine, carbocisteine alone, or standard care alone. The primary outcome was the number of exacerbations over 52 weeks; secondary outcomes included quality of life scores, time to next exacerbation, and safety. Results showed no statistically significant difference in exacerbation frequency at 52 weeks between the hypertonic saline, carbocisteine, combination therapy, and standard care groups. Secondary endpoints (lung function, time to first exacerbation, quality of life scores) and adverse event rates (including serious events) were also similar across groups. Neither primary nor secondary endpoints demonstrated clinical benefit [62].

As the first high-quality factorial-design RCT in this field, the CLEAR study—despite its neutral results—offers important clinical insight. It does not wholly negate the value of mucoactive drugs but indicates that pharmacologic intervention alone cannot replace active clearance techniques such as physiotherapy. For example, Targeted lung denervation (TLD) is a procedure in which a radiofrequency catheter is positioned and delivered into the main bronchi via bronchoscopy to suppress airway smooth muscle contraction, reduce airway mucus secretion, and modulate local inflammatory responses [117]. Furthermore, both bronchial rheoplasty and metered cryospray have been shown to effectively treat mucus obstruction by inducing necrosis and sloughing of epithelial and goblet cells in the airway mucosa [118, 119]. Future development of mucoactive therapies should shift from merely diluting sputum toward novel targets that restore mucociliary clearance mechanisms, thereby advancing bronchiectasis management from empiric sputum clearance toward precision-based strategies and laying groundwork for optimized disease control [62].

Studies on airway thickening in COPD remain relatively limited. The latest study involving 15 samples demonstrates that twice-daily treatment with the BGF pMDI for 52 weeks effectively alleviates small airway wall thickening. However, larger-scale, multicenter studies are still needed to further explore its value [121]. Furthermore, biologics such as tezepelumab have been shown to significantly improve small-to-medium airway remodeling in asthma, although this effect has not been validated in COPD cohorts [123]. In contrast, a recent single-center study in a COPD cohort has shown that mepolizumab can effectively improve imaging-manifested airway wall thickening in patients with COPD [145]. However, this finding should be considered emerging evidence requiring confirmation in larger trials. Although biologics have been shown to be beneficial in improving airway thickening in patients with COPD, the clinical selection of biologic agents is often based not only on imaging findings but also on the patient’s blood eosinophil count and the frequency of acute exacerbations. Therefore, before initiating treatment, a comprehensive assessment of the patient’s condition should be performed to guide clinical decision-making [146].

Conclusion

COPD is a highly heterogeneous syndrome characterized by significant variability in clinical presentation, pathological mechanisms, and disease progression among individuals. While spirometry remains the cornerstone of diagnosis, it falls short of capturing the full complexity and personalized nature of the disease. In recent years, advances in imaging techniques, particularly CT and MRI, have provided powerful, non-invasive tools for the precise assessment of structural and functional abnormalities in COPD.

This review systematically outlines four key imaging phenotypes of COPD: emphysema, small airway disease, airway mucus plugs, and airway wall thickening, and elaborates on their potential clinical significance as TTs. Imaging biomarkers such as the Parametric Response Map, total airway count, Pi10, and PiSlope may enable finer stratification of disease subtypes, assessment of severity, prediction of clinical outcomes, and could help inform personalized treatment decisions. For instance, imaging-based evaluation of emphysema distribution can assist in target selection for lung volume reduction surgery, while quantification of small airway disease may facilitate early intervention and optimization of drug delivery strategies.

Despite the promising potential of imaging in phenotyping and potentially guiding therapy for COPD, its integration into routine clinical practice remains challenging. Limitations include lack of technical standardization, concerns regarding radiation exposure, insufficient cost-effectiveness analyses, and complexities in integrating imaging data with conventional clinical parameters. In addition to these challenges, several practical obstacles impede the routine integration of imaging-defined treatable traits into clinical workflows. First, the lack of standardized acquisition protocols and post-processing methods—exemplified by the substantial methodological heterogeneity in calculating Pi10 and PiSlope—limits inter-study comparability and prevents the establishment of universal reference ranges. Second, advanced imaging techniques such as parametric response mapping require inspiratory-expiratory CT pairs, which increase radiation exposure, scanning time, and patient cooperation demands, while MRI-based methods like hyperpolarized gas imaging remain costly and confined to specialized centers. Third, artificial intelligence algorithms, despite showing promise, require rigorous external validation across diverse populations, scanner types, and disease severities to ensure generalizability before clinical deployment. Fourth, there is a gap in clinician training and awareness regarding the interpretation of quantitative imaging biomarkers and their integration into treatment decisions. Finally, regulatory, reimbursement, and informatics infrastructures have yet to accommodate the routine reporting of these biomarkers, further hindering their translation from research to practice.

Future research should focus on developing more intelligent, automated, and reproducible image analysis tools, promoting the integration of multimodal imaging with clinical and genomic data, and validating through prospective clinical trials whether imaging-guided treatment strategies truly improve long-term patient outcomes. Ultimately, imaging-based precision phenotyping holds the promise of shifting COPD management from a “one-size-fits-all” approach toward a new era of personalized medicine.

Authors’ contributions

W. X. drafted the initial manuscript, P.Y. assisted with chapter writing and created figures and tables. W.D. revised the manuscript. All authors read and approved the final manuscript.

Funding

This work was sponsored by the Program of National Natural Science Foundation of China (No.82000026), Chongqing medical scientific research project (Joint project of Chongqing Health Commission and Science and Technology Bureau, No.2025MSXM067), Natural Science Foundation of Chongqing, China (No. CSTB2025NSCQ-GPX1215), Postdoctoral Cultivation Fund of The First Affiliated Hospital of Chongqing Medical University, China (No.CYYY-BSHPYXM-202215), the Special Project of Postdoctoral Research Program of Chongqing (No. 2022CQBSHTB3004), China Postdoctoral Science Foundation (No. 2023MD744158), Chongqing Young and Middle-aged High-end Medical Talents Program(No.YXGD20255), Chengdu University of Traditional Chinese Medicine Undergraduate Research and Innovation Projects 2025–2026 (ky-2026018).

Data availability

No datasets were generated or analysed during the current study.

Declarations

Ethics approval and consent to participate

The data and results used in this paper were from published studies, and there were no ethical issues, so the approval of the ethics committee was not required.

Competing interests

The authors declare no competing interests.