Construction and Validation of A Nomogram to Identify Mucus Obstruction In Patients With Chronic Obstructive Pulmonary Disease

Summary

This study aimed to identify independent clinical predictors of computed tomography (CT)–detected small airway mucus plugs in patients with chronic obstructive pulmonary disease (COPD) and to construct and validate a nomogram for individualized risk prediction.

Abstract

Small airway mucus impaction in chest computed tomography (CT) is a clinically significant finding in chronic obstructive pulmonary disease (COPD), associated with accelerated pulmonary function decline, increased frequency of acute exacerbations, and higher susceptibility to respiratory infections. However, a validated predictive tool for identifying patients at risk of CT-detected mucus plugs is currently lacking. This study aimed to develop and validate a nomogram to predict small airway mucus obstruction in patients with COPD. We retrospectively enrolled 212 COPD patients from Shenzhen Second People’s Hospital (January 2021 to June 2022), of whom 47 had CT-confirmed mucus plugs (mucus plug group, MP) and 165 did not (non-mucus plug group, NMP). Univariate and receiver operating characteristic (ROC) analyses were used to identify candidate predictors. Multivariate logistic regression was conducted to construct the final predictive model, which was then transformed into a nomogram. Internal validation was performed using bootstrap sampling (1000 iterations). Bronchiectasis, chronic rhinosinusitis (CRS), body mass index (BMI), forced expiratory flow at 25–75% of predicted (FEF25–75%pred), residual volume-to-total lung capacity ratio (RV/TLC), and serum 25-hydroxyvitamin D [25(OH)D] were identified as independent risk factors for CT mucus plugs. The nomogram demonstrated excellent predictive value with an AUC of 0.9611. Calibration curves and decision curve analyses demonstrated good clinical utility. Bootstrap internal validation further supported the model’s predictive stability. This nomogram provides a practical, individualized tool to facilitate early identification and personalized management of COPD patients at risk of small-airway mucus obstruction.

Introduction

Chronic obstructive pulmonary disease (COPD) is characterized by persistent and largely irreversible airflow limitation. The World Health Organization states that it is projected to become the third leading cause of death globally by 2030¹. The disease primarily initiates in the small airways (airways with an internal diameter of less than 2 mm), which represent a fundamental site of COPD pathology. Structural and inflammatory changes in these regions often precede the emergence of clinical symptoms by several years, yet contribute substantially to the airflow obstruction. Pathological hallmarks of small airway disease in COPD include infiltration by inflammatory cells,^2,3,4 impairment of epithelial defense mechanisms^5,6 airway remodeling and fibrosis^7,8,9and the formation of mucus plugs (MP)^10,11.

Airway mucus plugs in COPD represent a pathological accumulation of mucus within the airway lumen, resulting in airflow limitation¹². Mucus plug formation is associated with a pro-inflammatory milieu, characterized by elevated eosinophil counts and upregulation of type 2 cytokine gene expression¹³. Excessive intraluminal mucus impairs oxygen diffusion and causes hypoxia in airway epithelial cells, creating conditions favorable to persistent bacterial colonization and recurrent lower respiratory tract infections¹⁴. These infections exacerbate disease severity and increase mortality risk¹⁵. Elevated airway mucus secretion has further been identified as a precursor to acute exacerbation events in COPD¹⁶. This highlights the critical need for early detection and a mechanistic understanding of the factors contributing to mucus plugs in patients with COPD.

A range of risk factors have been associated with airway mucus plug formation in chronic airway diseases, including viral infections^17,18, colonization by Pseudomonas aeruginosa^19,20 recurrent acute exacerbation episodes, impaired pulmonary function as measured by forced expiratory volume in 1 second (FEV1)²¹, smoking history²², elevated eosinophil peroxidase levels²³, intrabronchial mucin 5B (MUC5B) protein concentrations, and 25-hydroxyvitamin D (25(OH)D) levels, as well as infections attributable to mycoplasma and Aspergillus. species^24,25,26. Nevertheless, the specific risk profile for mucus plug development in COPD patients remains incompletely characterized, and the prognostic utility of individual risk factors in isolation is limited.

A multifactorial approach integrating several predictors may yield more clinically meaningful risk stratification. Nomograms have been widely applied across medical specialties, including oncology, cardiology, and pulmonology, to facilitate survival predictions, risk stratification, and therapeutic decision-making²⁷. They provide a nuanced, interpretable way to capture complex interactions among diverse clinical variables. Despite their broad utility, no validated nomogram exists to predict CT-detected mucus plugs in COPD patients. This study addresses this gap by identifying independent risk factors for mucus plug formation in COPD and developing a validated predictive nomogram to enable individualized risk assessment. Such a tool could be readily integrated into routine COPD management workflows, particularly in centers with access to HRCT imaging and spirometry, to support early targeted interventions and reduce the burden of exacerbations in at-risk patients.

Protocol

The present study was approved by the Ethics Committee of Shenzhen Second People’s Hospital (Protocol No. 20193357024). Informed consent was obtained from all participants or their legal representatives prior to enrollment.

Study population and methodology

This study was designed as a single-center, retrospective cohort study. Medical records of patients with a primary diagnosis of COPD at the Department of Respiratory Medicine, Shenzhen Second People’s Hospital, from January 2021 to June 2022 were reviewed. All adult patients (≥18 years) with a primary diagnosis of COPD were initially screened using International Classification of Diseases (ICD-10) coding and chart review from the hospital’s electronic medical record (EMR) system.

Inclusion criteria

(1) Confirmed diagnosis of COPD in accordance with the Global Initiative for chronic obstructive lung disease (GOLD) guidelines; (2) Availability of high-resolution computed tomography (HRCT) of the chest performed within one week of hospitalization; (3) Availability of complete spirometry and laboratory data; and (4) At least one year of follow-up data for acute exacerbation monitoring.

Exclusion criteria

(1) Active pulmonary infections (e.g., pneumonia or tuberculosis) at the time of HRCT imaging; (2) Coexisting lung malignancy; (3) Prior thoracic surgery with potential impact on airway anatomy; and (4) Missing critical clinical data or non-evaluable imaging due to motion artifacts. After applying these criteria, a final cohort of 212 patients was enrolled, comprising 47 patients in the mucus plug-positive (MP) group and 165 patients in the non-mucus plug (NMP) group. Representative HRCT images are illustrated in Figure 1. Patients in the NMP group (n = 165) served as internal controls, enabling statistical comparison of clinical characteristics, pulmonary function indices, and laboratory biomarkers between groups. All analyses were conducted on this internally controlled cohort to support hypothesis-driven model development.

Data collection

Data extraction followed a structured, sequential protocol. Demographic variables collected included age, sex, body mass index (BMI), and smoking status. Clinical history variables comprised COPD duration, acute exacerbation frequency, and comorbidities. Spirometry parameters retrieved including FEV1%, FEV1 to forced vital capacity(FVC), Vital capacity(VC), forced expiratory flow(FEF25–75%pred), Residual volume(RV), Total lung capacity(TLC), and the RV/TLC ratio. Laboratory indices included serum total immunoglobulin E (IgE), 25-hydroxyvitamin D(25(OH)D), serum calcium (Ca²⁺), phosphorus, Carbohydrate antigen (CA199), and fractional exhaled nitric oxide (FeNO), and conducting airway nitric oxide (CaNO). Comorbidity screening included sinusitis, asthma, bronchiectasis, fungal and bacterial colonization, and cardiovascular and metabolic diseases. All data was retrieved from the hospital’s electronic medical record (EMR) system. HRCT images were accessed from the hospital’s picture archiving and communication system (PACS) archive. Details of the software and equipment used in this study are provided in the Table of Materials. No physical reagents or laboratory materials were used; all analyses were performed using existing clinical and radiological data. All patient data were reviewed by two independent investigators. Missing data were handled using the ‘missForest’ non-parametric imputation method implemented in R, to minimize distortion in multivariate analyses.

HRCT diagnostic criteria for mucus plugs

All patients underwent HRCT using standardized institutional imaging protocols. Mucus plugs were defined radiologically on axial CT slices as identified as tubular or branching soft-tissue attenuation structures occupying an airway lumen, visible on at least two contiguous axial slices, consistent with published diagnostic criteria. Only cases with clearly demarcated, segmental or subsegmental airway opacities with soft tissue attenuation similar to soft tissue and not attributable to artifacts or bronchiectasis alone were labeled as mucus-plug positive. HRCT imaging was performed using a Siemens SOMATOM Definition AS (128-slice) CT scanner with the following acquisition parameters: slice thickness 1.0 mm, reconstruction interval 0.75 mm, and use of the B70f high-resolution kernel. Images were reviewed in standard lung window settings (window width : 1600 Hounsfield units [HU]; Window level: 600 HU. Two board-certified thoracic radiologists with over 8 years of experience independently reviewed all scans. Cases with interpretive discrepancies were resolved by consensus discussion. Diagnostic criteria were applied uniformly across all cases to ensure classification consistency.

Nomogram construction, evaluation, and validation

A nomogram was developed to predict CT-detected mucus plugs in COPD patients based on multivariate logistic regression results. The final model incorporated the following independent predictors: bronchiectasis, chronic rhinosinusitis (CRS), acute exacerbations (AE), BMI, FEF25–75%pred, RV/TLC ratio, and serum 25(OH)D levels. Each predictor is assigned a score on a horizontal points scale; individual scores are summed to yield a total score, which corresponds to a predicted probability of mucus plug presence on the output probability scale. The nomogram was subjected to internal validation via bootstrap resampling (1000 iterations) to assess predictive accuracy and discrimination using calibration curves (AUC and ROC).

Statistical analyses

All statistical analyses were performed using R version 4.1.2 and IBM SPSS Statistics version 25.0. Categorical data were expressed as frequencies and percentages; comparisons between groups were performed using the chi-squared test or Fisher’s exact test, as appropriate. Continuous data with normal distribution were expressed as mean ± standard deviation (SD) and compared using the independent samples t-test; non-normally distributed continuous data were expressed as median (interquartile range (IQR) and compared using the Mann-Whitney U test. Variables with P < 0.1 in the univariate logistic regression analysis were included in the model, consistent with standard practice in predictive model development. The R packages used were “rms”, “mstate”, “data.table”, “pROC”, “rmada”, “rio”, “boot”, and “missForest”. Nomogram construction was implemented using the lrm and nomogram functions from the rms package. ROC curves and AUC values were computed using the roc and auc functions from the pROC package. Calibration curves were generated with the calibrate function in RMS. Decision curve analysis (DCA) was performed using the decision curve function from the rmda package. Missing data imputation was performed using the missForest function. Bootstrap internal validation (1000 iterations) was conducted using the boot package. A fixed random seed (set.seed[240708] was applied at the start of the analysis to ensure reproducibility. A P-value of < 0.05 was considered statistically significant. The logistic regression model formula was:

glm(mucus_status ~ bronchiectasis + CRS + BMI + FEF25_75 + RV_TLC + VitD, family = "binomial")

Results

Baseline characteristics

The study comprised a cohort of 212 patients with COPD, divided into two groups: 47 with mucus plugs (MP) and 165 without mucus plugs (NMP). The occurrence of mucus plugs in this COPD population was found to be 28.33%. Statistical analysis, detailed in Table 1, identified significant differences between the MP and NMP groups in several key metrics. These included body mass index (BMI), the frequency of acute exacerbations (AE), prevalence of bronchiectasis and chronic rhinosinusitis, forced expiratory flow at 25–75% of pulmonary volume (FEF25–75pred%), the residual volume to total lung capacity ratio (RV/TLC), and serum levels of carbohydrate antigen 199 (CA199) and 25-hydroxyvitamin D (25(OH)D), each demonstrating a P-value less than 0.05. The COPD patients in the MP group had significantly higher AE, combined bronchiectasis, sinusitis, fungal infection, and CA199 index than the NMP group (P < 0.05), and significantly lower BMI, FEF 25–75 pred%, and RV/ TLC than the NMP group (P < 0.05). Table 1 presents the baseline demographic and clinical characteristics of the COPD cohorts, offering a detailed, quantitative overview of the study population. This table is crucial for highlighting the clinical and physiological differences between the MP and NMP groups among COPD patients, thereby laying the groundwork for further analysis and clinical interpretation.

Univariate logistic regression analysis

To identify potential predictors of mucus plug formation, we first conducted univariate logistic regression analyses for the clinical and radiological variables described above. Several factors demonstrated associations with mucus plug presence at a P < 0.1 threshold and were therefore selected for further evaluation. This inclusive criterion helped ensure that relevant variables were not excluded prematurely. These candidate predictors were subsequently subjected to ROC analysis and multivariate logistic regression to develop the final predictive model.

ROC analysis and optimal cutoff values

In this study, mucoid impaction was defined as the dependent variable. We selected eight variables showing statistically significant differences between the MP (Mucoid impaction positive) and NMP (Mucoid impaction negative) groups for receiver operating characteristic (ROC) curve analysis. The results of this analysis are methodically presented in Table 2. Furthermore, using ROC curve analysis, the optimal cutoff values for these variables were determined, with the findings thoroughly documented in Table 3. In this study, optimal cutoff points for variables were determined using the maximal Youden Index, as detailed in the table. Mucus plug-positive status was defined based on HRCT criteria: presence of soft-tissue density within the bronchial lumen occupying at least 50% of the airway diameter, present on at least two consecutive axial slices, and consistent with mucus rather than artifact or fluid. Cutoff values for continuous variables (e.g., FEF25–75, RV/TLC, vitamin D) were determined using ROC curve analysis. The optimal threshold for each was determined using the Youden index (sensitivity + specificity – 1), which identifies the threshold that simultaneously maximizes sensitivity and specificity. This identifies the value that maximizes sensitivity and specificity. These thresholds were used to transform variables into binary categories for multivariate logistic regression.

Multivariate Logistic regression analysis of MP

An advanced stepwise logistic regression analysis was performed, with the presence of mucus plugs as the dependent variable. The analysis used dichotomized predictors to identify independent risk factors. These findings are elaborated in Table 4. Using variables significant in the univariate analysis, the multivariate logistic regression model revealed statistically significant results. The analysis identified several independent risk factors for CT-detected mucus plugs in COPD patients. These included bronchiectasis, with an Odds Ratio (OR) and 95% Confidence Interval (CI) of 13.699 (4.256, 44.1); chronic rhinosinusitis, with an OR 95% CI of 7.291 (1.867, 28.467); body mass index, with an OR 95% CI of 0.17 (0.053, 0.547); Forced expiratory flow at 25–75% of pulmonary volume predicted (FEF25–75% pred), with an OR 95% CI of 0.091 (0.027, 0.307); residual volume to total lung capacity ratio (RV/TLC), with an OR 95% CI of 0.144 (0.038, 0.541); and serum 25-hydroxyvitamin D (25(OH)D) levels, with an OR 95% CI of 0.042 (0.011, 0.151) (P < 0.05). These findings are elaborated in Table 5.

Nomogram evaluation

The nomogram constructed in this study is a visual translation of the multivariate logistic regression model and serves as an individualized, interpretable risk estimation tool. Each predictor in the model is assigned a point value on a horizontal axis; these points are summed to yield a total score, which maps to a probability scale indicating the risk of mucus plug presence. This graphical interface allows clinicians to estimate patient-specific risk using routinely available clinical and imaging data. This approach follows previously validated frameworks, such as the pulmonary embolism nomogram proposed. Figure 1 depicts the column line diagram modeling, and Figure 2 displays a column-line diagram that was constructed to visualize the relative influence of predictive features in the nomogram model. Columns represent individual risk factors (e.g., bronchiectasis, CRS, BMI), while line heights indicate their contribution strength to the predicted probability of mucus plug presence. The diagram aids the interpretation of feature weightings and interactions. All values were generated from multivariate logistic regression output. No error bars or scale bars are applicable; the validation results, illustrated in Figure 3, demonstrate a significant concordance between predicted and actual occurrences of mucus plugs in COPD patients. The AUC in Figure 4 validates the model's accuracy. The pattern shown by the calibration curves in Figure 5 emphatically highlights the nomogram's reliable predictive value in a clinical setting, while Figure 6 emphasizes the accuracy of the reliability of the model through the sensitivity plot.

DATA AVAILABILITY:

All relevant raw data supporting the findings of this study have been submitted as the Supplementary Table.

Figure 1: Representative HRCT image of a COPD patient from the mucus plug-positive (MP) group demonstrating small airway mucus plug status. Yellow arrows indicate tubular soft-tissue attenuation structures occupying small airway lumens on contiguous axial slices, consistent with mucus plug formation. The image was acquired using a Siemens SOMATOM Definition AS (128-slice) CT scanner with lung window settings (width: 1,600 HU; level: −600 HU). Please click here to view a larger version of this figure.

Figure 2: Nomogram for predicting CT-detected small airway mucus plugs in COPD patients. Each predictor is represented on a horizontal axis with an assigned point value. Individual scores are summed to generate a total score, which maps to the predicted probability of mucus plug presence on the output scale. Predictors included: bronchiectasis, chronic rhinosinusitis (CRS), body mass index (BMI), FEF25–75%pred, RV/TLC ratio, and serum 25(OH)D levels. All values were derived from the multivariate logistic regression model. Please click here to view a larger version of this figure.

Figure 3: Receiver operating characteristic (ROC) curve of the nomogram. The red curve demonstrates the model’s discriminatory performance in distinguishing mucus plug-positive from mucus plug-negative COPD patients. The x-axis represents the false positive rate (1 − specificity), and the y-axis represents the true positive rate (sensitivity). The diagonal reference line represents a non-discriminating classifier. Please click here to view a larger version of this figure.

Figure 4: Calibration curve for internal validation of the nomogram. The x-axis represents the nomogram-predicted probability, and the y-axis represents the observed (actual) probability of mucus plug presence. Three curves are displayed: Apparent (dotted), Bias-corrected (solid), and Ideal (dashed). Bootstrap internal validation was performed with B = 1,000 repetitions (n = 212); mean absolute error = 0.035, indicating strong agreement between predicted and observed probabilities. Please click here to view a larger version of this figure.

Figure 5: Decision curve analysis (DCA) for the nomogram. Net clinical benefit (y-axis) is plotted against a range of high-risk threshold probabilities (x-axis) for three strategies: the nomogram (red), treat-all (blue), and treat-none (black). The nomogram demonstrates superior net benefit compared with default strategies across the clinically relevant threshold range. Please click here to view a larger version of this figure.

Figure 6: Bootstrap-validated receiver operating characteristic (ROC) curve of the nomogram. The black curve represents the mean ROC curve, and the red error bars indicate the variability across 1,000 bootstrap resampling iterations. The area under the curve (AUC = 0.9611; 95% CI: 0.9382–0.984) confirms the high discriminatory performance and predictive stability of the nomogram. Please click here to view a larger version of this figure.

Variable	MP (n = 47)	NMP (n = 165)	P-value
Age (years)	68 (65–78)	69 (64–74)	0.39
Male (%)	39 (82.98%)	139 (84.24%)	1
Female (%)	8 (17.02%)	26 (15.76%)	—
BMI (kg/m²)	20.76 (19.55–23.10)	23.03 (21.90–24.51)	<0.001
Disease duration (months)	10 (5–20)	10 (5–13)	0.06
Smoking (pack-years)	30 (0–40)	20 (0–40)	0.35
AE ≥2/year (%)	17 (36.17%)	19 (11.52%)	<0.001
Respiratory failure (%)	8 (17.02%)	25 (15.15%)	0.82
Bronchiectasis (%)	33 (70.21%)	38 (23.03%)	<0.001
Chronic rhinosinusitis (%)	18 (38.30%)	24 (14.55%)	<0.001
Fungal infection (%)	5 (10.64%)	5 (3.03%)	0.05
FEF25–75%	12.00 (9.40–18.71)	19.00 (13.27–29.30)	<0.001
RV/TLC (%)	45.51 (42.85–49.25)	48.68 (43.32–54.51)	0.02
CA199	26.82 (17.65–49.94)	13.86 (10.60–20.61)	<0.001
25(OH)D (ng/mL)	21.05 (18.49–23.40)	25.32 (23.66–27.74)	<0.001

Table 1: Baseline clinical and demographic characteristics of the study cohort. Comparison between the mucus plug-positive (MP) and mucus plug-negative (NMP) groups. Data are presented as n (%), mean ± SD, or median (IQR) as appropriate. Abbreviations: BMI, body mass index; AE, acute exacerbations; FEV1%, forced expiratory volume in 1 second percentage predicted; FEV1/FVC, FEV1-to-forced vital capacity ratio; FEF25–75%pred, forced expiratory flow at 25–75% predicted. Abbreviations; RV = residual volume; TLC = total lung capacity; RV/TLC = residual volume-to-total lung capacity ratio; IgE = immunoglobulin E; 25(OH)D = 25-hydroxyvitamin D; CA199 = carbohydrate antigen 199; FeNO = fractional exhaled nitric oxide; CaNO = conducting airway nitric oxide.

Variable	AUC	95% CI	P-value
25(OH)D	0.826	0.755–0.896	<0.001
BMI	0.737	0.652–0.821	<0.001
CA199	0.757	0.670–0.843	<0.001
Bronchiectasis	0.736	0.651–0.820	<0.001
FEF25–75%	0.716	0.632–0.800	<0.001
RV/TLC	0.616	0.535–0.697	0.015
AE	0.623	0.526–0.721	0.01
Chronic rhinosinusitis	0.619	0.522–0.716	0.013

Table 2: ROC analysis results for candidate predictor variables. Area under the curve (AUC) values are presented for eight variables demonstrating statistically significant differences between the MP and NMP groups, along with 95% confidence intervals, sensitivity, and specificity.

Variable	Cutoff	Sensitivity	Specificity	Youden Index
BMI	21.11	0.842	0.617	0.459
25(OH)D	23.06	0.806	0.745	0.551
RV/TLC	49.82	0.473	0.787	0.26
FEF25–75%	15.35	0.679	0.702	0.381
CA199	17.08	0.809	0.685	0.494
Bronchiectasis	0.5	0.702	0.77	0.472
AE	0.5	0.362	0.885	0.247
Chronic rhinosinusitis	0.5	0.383	0.855	0.238

Table 3: Optimal cutoff values for continuous predictors. Threshold values were determined by the Youden index (sensitivity + specificity − 1) from ROC curve analysis. Variables were dichotomized at these thresholds prior to entry into multivariate logistic regression.

Variable	OR	95% CI	P-value
BMI	0.116	0.056–0.239	<0.001
25(OH)D	0.082	0.039–0.177	<0.001
FEF25–75%	0.201	0.099–0.406	<0.001
RV/TLC	0.301	0.141–0.646	0.002
CA199	7.109	3.403–14.852	<0.001
Bronchiectasis	7.878	3.825–16.226	<0.001
AE	4.354	2.030–9.341	<0.001
Chronic rhinosinusitis	3.647	1.757–7.568	0.001

Table 4: Univariate logistic regression analysis of candidate predictors. Results are presented as odds ratios (OR) with 95% confidence intervals (CI) and corresponding P-values. Variables with P < 0.1 were selected for inclusion in the multivariate logistic regression model.

Variable	β	OR	95% CI	P-value
Bronchiectasis	2.617	13.699	4.256–44.100	<0.001
Chronic rhinosinusitis	1.987	7.291	1.867–28.467	0.004
BMI	-1.771	0.17	0.053–0.547	0.003
FEF25–75%	-2.397	0.091	0.027–0.307	<0.001
RV/TLC	-1.941	0.144	0.038–0.541	0.004
25(OH)D	-3.179	0.042	0.011–0.151	<0.001

Table 5: Stepwise multivariate logistic regression analysis identifying independent risk factors for CT-detected mucus plugs. Results are presented as odds ratios (OR) with 95% confidence intervals (CI) and P-values. P < 0.05 was considered statistically significant.

Discussion

In this study, the prevalence of CT-detected mucus plug formation among hospitalized COPD patients was 22.16%, consistent with estimates reported in prior literature²⁷. Mucus plugs in COPD are clinically significant due to their association with accelerated pulmonary function decline, increased acute exacerbation frequency, and higher mortality risk²⁸. Despite this, a validated predictive tool for identifying at-risk patients was previously lacking. This analysis identified bronchiectasis, chronic rhinosinusitis (CRS), BMI, FEF25–75%pred, RV/TLC, and 25(OH)D as independent risk factors for CT-detected mucus plugs, and these were integrated into a nomogram with excellent discriminatory performance.

Bronchiectasis demonstrated the strongest independent association with mucus plug formation (OR = 13.70), consistent with its established role in impairing mucociliary clearance and promoting mucus stasis. These findings support the conceptualization of bronchiectasis-COPD overlap (BCO) as a distinct clinical phenotype with heightened susceptibility to small airway obstruction²⁹. Bronchiectasis was radiologically detected in 24.5% of COPD patients in a prior study²⁹, and patients with concurrent disease exhibited more extensive airway involvement, including air trapping and peribronchial wall thickening. The independent association of CRS with mucus plug risk likely reflects the unified airway hypothesis, in which upper- and lower-airway inflammatory processes are mechanistically linked. Regarding BMI, patients hospitalized for acute COPD exacerbations with lower BMI had elevated sputum mucin and neutrophil elastase levels, suggesting that nutritional deficiency may potentiate mucus hypersecretion, particularly in advanced disease.

FEF25–75%pred is a sensitive spirometric marker of small airway obstruction, and its inverse association with mucus plug formation is consistent with published evidence linking CT mucus impaction to diminished small airway expiratory flow³⁰. A cross-sectional study demonstrated a significant correlation between luminal mucus scoring, lung function parameters, and health-related quality of life in COPD patients. Another study involving 500 participants reported a 22% prevalence of CT mucus impaction, with higher global initiative for chronic obstructive lung disease (GOLD) stage patients exhibiting greater impaction burden and lower FEV1 and FEF25–75% values; notably, 73% of patients with initial CT mucus obstruction retained it after five years. A further study demonstrated that FEV1 independently predicted luminal mucus scoring in hospitalized patients with acute exacerbations of COPD (AECOPD) (R2 = 0.348, F = 18.960, P < 0.001)³¹. Elevated RV/TLC reflects gas trapping, a physiological consequence of small airway disease, corroborating the role of dynamic hyperinflation in mucus plug pathogenesis³².

Vitamin D plays a critical role in pulmonary immune defense, including regulation of cytokine production, enhancement of macrophage phagocytosis, and attenuation of inflammatory responses³³. Prior studies have demonstrated associations between low serum 25(OH)D levels and worse lung function³⁴, or COPD severity³⁵. These findings of significantly lower 25(OH)D levels in the MP group are consistent with this body of evidence and underscore the potential role of vitamin D supplementation as a modifiable intervention target. Vitamin D deficiency has been associated with increased respiratory bacterial colonization^36,37, impaired ciliary clearance through alterations in intracellular and extracellular calcium homeostasis, and heightened susceptibility to respiratory infection. It was also shown that increased mortality in male patients with mild-to-moderate COPD is associated with significantly lower serum 25(OH)D levels^38,39.

Unlike established COPD prognostic tools such as the BODE index (body mass index, airflow obstruction, dyspnea, exercise capacity) or the ADO score (age, dyspnea, airflow obstruction)- which incorporate systemic clinical parameters and spirometric measurements (lung function indices derived from spirometry, including FEV1, FVC, and derived ratios) to predict outcomes such as exacerbation risk or mortality—this model specifically targets the radiological presence of small airway mucus plugs. This represents a distinct pathological feature with independent clinical implications that is not addressed by existing risk tools. This nomogram, therefore, provides added value for airway-level phenotyping in COPD populations. With further multicenter external validation, the model could be embedded in radiology reporting platforms or electronic health record (EHR) systems to flag high-risk patients for early mucolytic therapy, airway clearance therapies, or bronchoscopic intervention.

This study also demonstrates the value of integrating multiple biomarker domains, serological, functional, radiological, and clinical history, into a single predictive framework. An animal-model pharmacological study demonstrated that tetrandrine significantly reduces excessive MUC5AC production and suppresses expression of TNF-α, IL-6, IL-8, and IL-17A in a lipopolysaccharide-induced mucus hypersecretion model⁴⁰, suggesting candidate therapeutic pathways. Population-level evidence from the Copenhagen cohort established a strong association between impaired pulmonary function, chronic mucus hypersecretion, and both all-cause and COPD-specific mortality⁴¹ while the COPD Gene study confirmed that CT-detected luminal obstruction correlates with airflow limitation, diminished quality of life, and emphysematous phenotypes⁴².

In conclusion, this study identifies bronchiectasis, chronic rhinosinusitis, BMI, FEF25–75% pred, RV/TLC, and serum 25(OH)D as independent risk factors for CT-detected small airway mucus plugs in COPD patients and presents a validated nomogram with high predictive accuracy (AUC = 0.96), strong calibration, and demonstrated clinical utility. The model is distinguished by its integration of diverse predictor domains, its interpretable graphical format, and its reliance on routinely available clinical data. It offers potential for future integration into COPD care workflows and EHR systems to support individualized, data-driven decision-making.

This retrospective single-center study is subject to inherent selection bias, and the relatively small sample size (n = 212) limits the statistical power for subgroup analyses. The nomogram currently lacks external validation across independent patient populations and imaging systems, which is a critical prerequisite for broad clinical adoption. The study was conducted at a single Chinese tertiary center, and generalizability to other ethnic populations or healthcare settings requires investigation. Future studies should pursue prospective multicenter validation, consider machine learning approaches (such as random forests or gradient boosting) to enhance predictive performance and automate feature selection, and assess the longitudinal predictive value of the model for monitoring mucus plug dynamics and treatment response over time.

Materials

List of materials used in this article

Name	Company	Catalog Number	Comments
HRCT Scans	Shenzhen Second People's Hospital		Used for diagnosing small airway mucus obstruction in COPD patients
SPSS 25.0 Software	1BM		Statistical software used for data analysis, including t-tests and logistic regression.
R Software (Packages: mms, mstate, etc.)	R Foundation for Statistical Computing		Used for statistical analysis and model validation, including calculation of the C-index.
Electronic Medical Record System	Shenzhen Second People's Hospital		Data source for clinical and laboratory variables, including patient history and diagnostic parameters.
Logistic Regression Equation	Custom (Applied via SPSS and R)		Used to screen for independent risk factors related to small airway mucus obstruction in COPD patients.