Assessment of CT quantitative characteristics derived from CT in connective tissue disease-associated interstitial lung disease Authors: Yuxin Cheng, Rongrong Fan, Yueze Li, Jiaming Zheng, Li Fan, Shiyuan Liu, Yu Guan, Yi Xiao Published: 2026 ABSTRACT AbstractBackgroundConnective tissue disease-associated interstitial lung disease (CTD-ILD) is a severe complication, yet early objective detection of pulmonary structural and microvascular alterations remains challenging. This study aimed to quantitatively compare pulmonary structural alterations between connective tissue disease (CTD) patients with and without interstitial lung disease (ILD) and to identify independent computed tomography (CT)-derived discriminators for ILD diagnosis by means of quantitative computed tomography (QCT).MethodsFifty-one CTD patients with ILD (median age, 51 years; 66.7% female) and thirty-three CTD patients without ILD (median age, 33 years; 69.7% female) who underwent paired inspiratory and expiratory non-contrast chest computed tomography (CT) scans and pulmonary function testing were retrospectively collected. The lung density, functional small airway and pulmonary vessel parameters were analyzed using computer software. Differences in these CT quantitative parameters between the patients with CTD-ILD and those with CTD without ILD were compared using Mann-Whitney U tests. Furthermore, univariable and multivariable logistic regression analyses were used to establish nomograms to identify the independent predictors associated with the presence of ILD in patients with CTD. The calibration curve evaluates the predictive accuracy, while the decision curve analysis (DCA) evaluates clinical applicability.ResultsThe forced vital capacity (FVC), forced expiratory volume in the first second (FEV1), diffusing capacity of the lungs for carbon monoxide corrected for alveolar volume (DLCO/VA) and diffusing capacity of the lungs for carbon monoxide (DLCO) of the CTD with ILD group were significantly lower than those of the CTD without ILD group (all P<0.05). CTD with ILD group had higher percentages of high-attenuation area on inspiratory and expiratory CT (HAA%-IN and HAA%-EXP) (both P<0.05), and lower percentage of low-attenuation area on expiratory CT (LAA%-EXP), functional air trapping (fAT) volume, functional small airway disease (fSAD) volume and fSAD volume ratio (all P<0.001). Additionally, pulmonary vascular parameters including No. vessels, No. vessels cross-sectional area (CSA) <5 mm2 at 6, 12 and 24 mm depth from the pleural surface, BV1, BV5, BV10 and total blood volume (TBV) were significantly decreased in CTD-ILD patients (all P<0.001). Multivariate Ridge regression identified age, sex, DLCO%, FEV1%, EXP HAA%, fSAD ratio, and No. vessels_12mm as independent diagnostic predictors associated with the presence of ILD (all P<0.05). The combined diagnostic model achieved an area under the curve (AUC) of 0.963 [95% confidence interval (CI): 0.920–0.998], with an accuracy of 0.893, sensitivity of 0.941 and specificity of 0.818. The calibration curve demonstrated high consistency between predicted probabilities and actual outcomes.ConclusionsQCT can serve as a crucial and highly promising imaging biomarker for the diagnosis and assessment of CTD-ILD.Keywords: Interstitial lung disease (ILD), connective tissue diseases (CTDs), pulmonary fibrosis, quantitative computed tomography (QCT) Highlight box Key findings • A combined diagnostic model integrating quantitative computed tomography (QCT) parameters—specifically high attenuation area, functional small airway disease ratio, and pulmonary vessel counts—demonstrated exceptional performance in identifying connective tissue disease-associated interstitial lung disease (CTD-ILD), achieving an area under the curve of 0.963. What is known and what is new? • Interstitial lung disease (ILD) is a severe complication of connective tissue disease (CTD). However, early and objective detection of subtle pulmonary structural and microvascular changes remains a challenge for conventional visual computed tomography assessment. • This study identifies that CTD-ILD patients exhibit significantly reduced functional small airway volume and peripheral pulmonary vessel loss. By utilizing automated software, this research establishes a novel multivariate diagnostic model that combines these QCT biomarkers with pulmonary function tests to independently predict the presence of ILD. What is the implication, and what should change now? • QCT parameters serve as objective, non-invasive and highly promising imaging biomarkers for the diagnosis of CTD-ILD. Clinical practice should move toward incorporating computer-aided quantitative analysis of lung density, small airways, and vasculature. Introduction Connective tissue diseases (CTDs) are a group of chronic inflammatory disorders characterized by abnormal activation and dysregulation of the immune system. The lungs are one of the most frequently involved target organs in CTD, which can lead to various pulmonary complications (1). Among these, connective tissue disease-associated interstitial lung disease (CTD-ILD) is a key factor influencing patient prognosis. In the early stages of the disease, inflammatory cell infiltration primarily affects the bronchioles and alveolar walls, leading to fibrotic changes, bronchiolar obstruction, and thickening of the alveolar septa. These pathological alterations impair gas exchange, resulting in restricted lung function and reduced ventilation efficiency. Prolonged hypoxia further induces endothelial dysfunction, damages the pre-capillary pulmonary vasculature, and ultimately promotes pulmonary vascular remodeling, potentially progressing to pulmonary hypertension (PH) (2-4). On high-resolution computed tomography (HRCT) of the chest, CTD-ILD typically presents with reticular opacities, traction bronchiectasis, and honeycombing predominantly in the lower lobes, consistent with the imaging features of nonspecific interstitial pneumonia (NSIP). Although certain imaging and pathological manifestations are recognized, the exact pathogenesis and progression of CTD-ILD remain incompletely understood. Particularly in CTD patients without apparent interstitial changes in the lungs, early microstructural and functional alterations within the lungs are still unclear. There is also a lack of objective, quantitative indicators to delineate the continuum from CTD without ILD to established CTD-ILD. Current joint guidelines from the European Respiratory Society and the European Alliance of Associations for Rheumatology (ERS/EULAR) recommend HRCT combined with pulmonary function tests (PFTs) as the primary approach for screening and long-term monitoring of CTD-ILD (5). However, both methods have limitations. PFT results are significantly influenced by patient cooperation and operational standardization, leading to considerable variability. As a global assessment of lung function, PFTs also exhibit a certain lag in detecting early pathological changes. HRCT evaluation currently relies mainly on visual qualitative judgments and semi-quantitative scoring, which is not only time-consuming but also suffers from inter-observer variability and limited sensitivity to early disease and subtle progression. Therefore, there remains a need for ideal tools enabling early detection and accurate assessment of ILD. The development of objective, quantitative imaging analysis methods represents an important direction for enhancing early diagnosis and prognostic evaluation capabilities. Quantitative computed tomography (QCT) has emerged as a powerful non-invasive tool. Initially developed and extensively validated for obstructive airway diseases such as chronic obstructive pulmonary disease (COPD) and asthma, density-based QCT metrics provide distinct biological insights. Specifically, low-attenuation areas (LAA) and functional small airway disease (fSAD) were developed to quantify the extent of parenchymal destruction (emphysema) and expiratory air trapping, respectively, by measuring voxels below predefined low-density thresholds (e.g., −950 and −856 HU) (6,7). Conversely, high-attenuation areas (HAA) capture voxels with abnormally increased density, making them highly effective for quantifying ground-glass opacities, reticulations, and fibrotic burden in ILD (8). Furthermore, recent advancements in QCT automated segmentation allow for the precise measurement of small pulmonary vessels, serving as a surrogate marker for microvascular pruning and remodeling (9). Although these individual metrics are well-established, their comprehensive, multi-domain application in characterizing CTD-ILD remains insufficiently explored (10). QCT has the potential to detect pulmonary structural alterations at a stage earlier than traditional radiological assessments by experienced radiologists. The purpose of this study is to investigate the quantitative pulmonary manifestations in CTD-ILD and to identify independent risk factors for the diagnosis of ILD in CTD patients utilizing QCT parameters. This research aims to enhance early diagnostic capabilities and contribute to the development of targeted therapeutic strategies for CTD-ILD. We present this article in accordance with the TRIPOD reporting checklist (available at https://jtd.amegroups.com/article/view/10.21037/jtd-2025-1-2705/rc). Methods Subjects This retrospective study enrolled patients with CTD at their first visit to the Rheumatology Department of The Second Affiliated Hospital of Naval Medical University from November 2023 to October 2024. Participants were diagnosed with CTD according to the criteria established by the rheumatology society (11-13). The inclusion criteria were as follows: these patients with CTD underwent PFT in The Second Affiliated Hospital of Naval Medical University and completed inspiratory and expiratory CT scans within 2 weeks. Exclusion criteria were: (I) age below 18 years; (II) concomitant pulmonary diseases, such as infection, pulmonary embolism, pneumothorax, atelectasis, pleural effusion, COPD or ILD caused by pneumoconiosis, inhalation of organic matter and immunization therapy; (III) other concomitant diseases that could interfere with respiratory function and/or quality of life, such as severe cerebrovascular disease, severe heart failure caused by cardiovascular disease, severe renal disease. The inclusion and exclusion flowchart is shown in Figure 1. The demographic and clinical characteristics of the population, including basic information [age, gender, body mass index (BMI)] were recorded. Flow chart of study inclusion, CT, computed tomography; CTD-ILD, connective tissue disease-associated interstitial lung disease. Patients were categorized into the CTD with ILD group or CTD without ILD group based on the presence or absence of ILD imaging features on the inspiration images. Two thoracic radiologists independently assessed the images according to the ATS/ERS/JRS/ALAT guidelines for interstitial pneumonia [2022] (14), and the Fleischner Society glossary of thoracic imaging terms [2024] (15); discrepancies were resolved by consensus. The imaging evaluators were blinded to the results of pulmonary function and CT quantitative assessments. The study was approved by the Institutional Review Board of The Second Affiliated Hospital of Naval Medical University (No. 2018L028), and the study was registered in the Chinese Clinical Trials Registry (http://www.chictr.org.cn/index.aspx; ChiCTR2000035283). The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. Given that this is a retrospective study and all data were extracted from the imaging data system, the requirement for informed consent was waived by the Institutional Review Board. PFT The spirometer (Master Screen Body; Germany) was used to perform PFT. Lung function was measured according to GOLD guidelines (16). Forced expiratory volume in the first second (FEV1), forced vital capacity (FVC), FEV1 to FVC ratio (FEV1/FVC), diffusing capacity of the lungs for carbon monoxide corrected for alveolar volume (DLCO/VA), and DLCO were measured as percentages of predicted values. CT scanning All the subjects are screened with a craniocaudal CT with the subject lying supine and both arms raised. In the supine position, a dual-source CT system (SIEMENS SOMATOM Force, Siemens Healthcare, Forchheim, Germany) was used to perform chest scans in both full inspiration and expiration states of the participants. Subjects underwent breath-hold training before CT scanning. No contrast-enhanced volumetric chest CT scanning was performed at the end of inspiration and expiration from the thoracic inlet to the diaphragm. The acquisition and reconstruction parameters for chest CT scanning are as follows: collimation 192×0.6 mm, tube energy 90/Sn150 kV, CARE Dose4D (Siemens Healthcare) was on, and a lung convolution kernel (Bl57) at a slice thickness of 1 mm and an increment of 0.7 mm. Field of view (FOV) 387 mm × 387 mm, matrix 512×512, An advanced modeled iterative reconstruction algorithm (ADMIRE, Siemens) at a level of 3 was applied. Image analysis The parameters of the lung density, functional small airway and pulmonary small vessels were analyzed with commercial software (A-VIEW, Suhai Alderi Information Technology Ltd., Dubai, UAE). The analysis provided the following metrics: total lung volume (in cc), percentages of low-attenuation area on inspiratory and expiratory CT (LAA%-IN and LAA%-EXP), percentages of high-attenuation area on inspiratory and expiratory CT (HAA%-IN and HAA%-EXP), volumes and volume ratios of functional small airway disease (fSAD volume and fSAD volume ratio) and functional air trapping (fAT volume and fAT volume ratio), as well as pulmonary vascular metrics including the number of small vessels (No. vessels), number of vessels with cross-sectional area (CSA) <5 mm2 at various depths, total blood volume (TBV), and blood vessel volumes stratified by CSA (BV1, BV5, BV10). Statistical analysis All statistical analyses were performed using commercially available software (SPSS version 27.01; IBM SPSS Statistics, Armonk, NY, USA, and R version 4.5.1; R Core Team, Vienna, Austria). Patient characteristics were reported as mean (standard deviation) for normally distributed data, median (interquartile range) for non-normally distributed data, or percentages for categorical variables, as appropriate. For key variables, 95% confidence intervals (CIs) were provided. Group comparisons were conducted using χ2 tests or Mann-Whitney U tests depending on data distribution and type. Univariate and subsequent multivariate logistic regression models were developed to identify independent predictors for CTD-ILD. To rigorously adjust for baseline imbalances and prevent model overfitting given the modest sample size, variables with P<0.05 from the univariate analysis were entered into a penalized machine learning framework. Specifically, least absolute shrinkage and selection operator (LASSO) regression with 10-fold cross-validation was employed for robust feature selection and to further eliminate multicollinearity. Subsequently, a Ridge regression (L2 regularization) model was constructed using the LASSO-selected variables to establish the final multidimensional diagnostic model. The diagnostic performance of the final model was evaluated using the area under the receiver operating characteristic (ROC) curve (AUC), accuracy, sensitivity, and specificity. Calibration curves were plotted to assess the model’s goodness-of-fit, and the DeLong test was used to compare the AUCs between different models or individual indicators. Results Subjects’ demographic and clinical characteristics The basic clinical and demographic features of the CTD-ILD group and the CTD without ILD group are summarized in Table 1. There were significant differences in BMI, FVC%, FEV1%, DLCO/VA and DLCO% between the two groups (all P<0.05), while no difference was found in FEV1/FVC and smoking history. The FVC%, FEV1%, DLCO/VA and DLCO% of the CTD without ILD group were significantly higher than those of the CTD with ILD group (all P<0.01). Meanwhile, we observed that the age of the CTD without ILD group was lower than that the CTD with ILD group. In terms of clinical characteristics, significant differences were observed in disease duration and CTD subtypes between the two groups. The mean disease duration was significantly shorter in the CTD-ILD group compared to the CTD without ILD group (6.97±5.54 vs. 11.77±8.16 years, P=0.005). Furthermore, the distribution of CTD subtypes differed markedly (P<0.001): the CTD without ILD group was predominantly composed of patients with systemic lupus erythematosus (SLE) (18 cases, 54.55%), whereas systemic sclerosis (SSc) constituted the vast majority of the CTD-ILD group (39 cases, 76.47%). Other subtypes included dermatomyositis, rheumatoid arthritis, and Sjögren’s syndrome. Regarding the radiological patterns of ILD on HRCT, the vast majority of the 51 CTD-ILD patients exhibited a NSIP pattern (n=48, 94.1%). Other patterns included usual interstitial pneumonia (UIP) (n=1, 1.9%), organizing pneumonia (OP) (n=2, 3.9%). Table 1Demographic and clinical characteristics between the CTD patients with and without ILDVariableCTD without ILD (n=33)CTD-ILD (n=51)PFemale23 (69.7)34 (66.7)0.77Weight (kg)57.39±6.6860.57±12.130.18Age (years)33.67±14.3351.78±13.22<0.001BMI (kg/m2)21.84±2.4123.32±3.430.03Smoking history5 (15.5)16 (31.4)0.09FVC (%)90.42±11.9571.48±15.72<0.001FEV1 (%)90.95±12.1273.89±15.38<0.001FEV1 to FVC ratio (%)85.89±8.6783.72±5.470.21DLCO (%)65.80±15.2343.86±15.36<0.001DLCO/VA (%)57.25±12.5148.24±14.440.004Disease duration (years)11.77±8.166.97±5.540.005CTD subtypes<0.001 Sjögren’s syndrome2 (6.06)1 (1.96) Rheumatoid arthritis5 (15.15)3 (5.88) Dermatomyositis7 (21.21)7 (13.73) Systemic lupus erythematosus18 (54.55)1 (1.96) Systemic sclerosis1 (3.03)39 (76.47)Patterns of interstitial lung disease UIP–1 (1.9) NSIP–48 (94.1) OP–2 (3.9)Data are shown as n (%) or mean ± standard deviation. BMI, body mass index; CTD, connective tissue disease; DLCO, diffusing capacity of the lungs for carbon monoxide; DLCO/VA, diffusing capacity of the lungs for carbon monoxide divided by alveolar volume; FEV1, forced expiratory volume in the first second; FVC, forced vital capacity; ILD, interstitial lung disease; NSIP, nonspecific interstitial pneumonia; OP, organizing pneumonia; UIP, usual interstitial pneumonia. Demographic and clinical characteristics between the CTD patients with and without ILD Data are shown as n (%) or mean ± standard deviation. BMI, body mass index; CTD, connective tissue disease; DLCO, diffusing capacity of the lungs for carbon monoxide; DLCO/VA, diffusing capacity of the lungs for carbon monoxide divided by alveolar volume; FEV1, forced expiratory volume in the first second; FVC, forced vital capacity; ILD, interstitial lung disease; NSIP, nonspecific interstitial pneumonia; OP, organizing pneumonia; UIP, usual interstitial pneumonia. Differences in parameters of lung density, functional small airway and pulmonary small vessels between the CTD patients with and without ILD In comparison to CTD without ILD, CTD with ILD had higher HAA%-IN and HAA%-EXP (both P<0.001). While the LAA%-EXP was significantly lower in CTD-ILD compared with that in CTD without ILD (P<0.01). Both fSAD volume and fSAD volume ratio were lower in CTD with ILD compared with CTD without ILD (both P<0.001). The fAT volume was also lower in CTD with ILD compared to CTD without ILD (P<0.001). Furthermore, there were significant differences for all the pulmonary vessels parameters between CTD with ILD and CTD without ILD (all P<0.001) (Table 2). The No. vessels, No. vessels CSA <5 mm2 at 6, 12 and 24 mm depth from the pleural surface, BV1, BV5, BV10 and TBV were significant lower in the CTD with ILD compared with those in the CTD without ILD (Figures 2-4). Table 2Differences in parameters of lung density, functional small airway and pulmonary vessels among normal, the CTD patients with and without ILDVariableCTD without ILD (n=33)CTD with ILD (n=51)PParameters of lung density LAA%-IN (%)3.04 (1.03, 6.08)2.35 (1.43, 5.58)0.98 HAA%-IN (%)3.93 (3.66, 4.50)7.12 (5.36, 9.82)<0.001 LAA%-EXP (%)16.49 (10.28, 34.69)7.03 (3.19, 14.86)<0.001 HAA%-EXP (%)9.46 (6.18, 16.59)21.10 (15.66, 28.69)<0.001Parameters of functional small airway fAT volume (cc)863.20 (322.60, 2461.22)264.74 (108.88, 596.94)<0.001 fAT ratio (%)19.30 (9.51, 44.93)8.09 (3.71, 16.28)0.001 fSAD volume (cc)770.89 (197.85, 1,622.46)170.23 (68.46, 416.43)<0.001 fSAD volume ratio (%)14.99 (6.44, 39.92)5.56 (2.59, 11.25)<0.001Parameters of pulmonary vessels No. vessels-6 mm (n)2,067.00 (1,728.00, 2,691.00)1,169.00 (939.50, 1,711.50)<0.001 No. vessels-12 mm (n)1161.00 (914.00, 1327.00)702.00 (593.50, 866.00)<0.001 No. vessels-24 mm (n)317.00 (233.00, 404.00)163.00 (126.50, 227.50)<0.001 No. vessels CSA <5 mm2-6 mm (n)1,863.00 (1,607.00, 2,394.00)948.00 (789.50, 1,452.50)<0.001 No. vessels CSA <5 mm2-12 mm (n)735.00 (612.00, 829.00)479.00 (377.50, 625.50)<0.001 No. vessels CSA <5 mm2-24 mm (n)143.00 (110.00, 193.00)81.00 (65.00, 114.75)<0.001 BV1 (cc)228.04 (183.60, 283.11)143.48 (131.36, 195.08)<0.001 BV5 (cc)11.03 (9.51, 12.65)8.13 (6.67, 10.15)<0.001 BV10 (cc)115.91 (91.09, 143.17)75.47 (68.11, 103.33)<0.001 TBV (cc)184.76 (146.67, 226.29)118.54 (109.80, 162.70)<0.001Data are presented as M (Q1, Q3). BV1, blood vessel volume in vessels with cross-sectional area <1 mm2; BV5, blood vessel volume in vessels with cross-sectional area <5 mm2; BV10, blood vessel volume in vessels with cross-sectional area <10 mm2; CSA, cross-sectional area; CTD, connective tissue disease; fAT, functional air-trapping voxels; fSAD, functional small airway disease; HAA, high-attenuation area; ILD, interstitial lung disease; LAA, low-attenuation area; M, median; No. vessels, the number of small pulmonary vessels; No. vessels CSA <5 mm2, the number of pulmonary vessels with a cross-sectional area less than 5 mm2; Q1, 1st quartile; Q3, 3rd quartile; TBV, total blood vessel volume. Differences in parameters of lung density, functional small airway and pulmonary vessels among normal, the CTD patients with and without ILD Data are presented as M (Q1, Q3). BV1, blood vessel volume in vessels with cross-sectional area <1 mm2; BV5, blood vessel volume in vessels with cross-sectional area <5 mm2; BV10, blood vessel volume in vessels with cross-sectional area <10 mm2; CSA, cross-sectional area; CTD, connective tissue disease; fAT, functional air-trapping voxels; fSAD, functional small airway disease; HAA, high-attenuation area; ILD, interstitial lung disease; LAA, low-attenuation area; M, median; No. vessels, the number of small pulmonary vessels; No. vessels CSA <5 mm2, the number of pulmonary vessels with a cross-sectional area less than 5 mm2; Q1, 1st quartile; Q3, 3rd quartile; TBV, total blood vessel volume. Quantitative analysis of lung density between CTD with and without ILD. (A,B) The HAA and LAA on inspiratory CT scans for CTD-ILD (left) and CTD without ILD (right). The HAA is depicted in blue with a threshold between −600 and −250 HU, while the LAA is shown in green with a threshold of less than −950 HU. The patient with CTD-ILD has a higher proportion of HAA and LAA. CT, computed tomography; CTD-ILD, connective tissue disease-associated interstitial lung disease; HAA, high-attenuation area; HU, Hounsfield Unit; LAA, low-attenuation area. Functional small airway disease analysis between CTD with and without ILD. (A,C) show a patient with CTD-ILD, while (B,D) show a patient with CTD but without ILD. (A) and (B) show air-trapping analysis for both patients, with orange and yellow areas indicating regions of air-trapping. It is noted that the patient with CTD without ILD has a larger fAT%, with values of 6% vs. 12%. (C) and (D) present Paramap analysis for both patients, with yellow areas indicating regions of SAD. The patient with CTD without ILD has a larger fSAD%, with values of 4% vs. 9%. CTD-ILD, connective tissue disease-associated interstitial lung disease; fAT%, functional air-trapping percentage; fSAD, functional small airway disease; SAD, small airway disease. Vascular trees between CTD with and without ILD, color-coded by vessel cross-sectional area, for a subject with CTD-ILD (A) and a CTD without ILD subject (B). It is observed that the patient with CTD-ILD has a lack of red areas indicating smaller vessels. CTD-ILD, connective tissue disease-associated interstitial lung disease. The model construction and performance for the identification of CTD patients with ILD To eliminate multicollinearity and select the most robust features from the candidate variables, a LASSO regression was initially applied, identifying 10 potential parameters: DLCO%, FVC%, FEV1%, age, sex, IN LAA%, EXP LAA%, EXP HAA%, fSAD ratio, and No. vessels_12mm. Subsequently, a rigorous bootstrap resampling procedure (n=1,000 iterations) was performed to evaluate the stability of these variables and construct the final multidimensional diagnostic model. Seven variables emerged as significant independent diagnostic indicators for CTD-ILD (all P<0.05). Notably, while FVC% was initially screened, it was excluded in the final robust model due to its strong collinearity with FEV1%. The final retained indicators included demographic factors: age [odds ratio (OR) =1.527, 95% confidence interval (CI): 1.214–1.920, P<0.001] and sex (OR =1.248, 95% CI: 1.049–1.485, P=0.01); physiological pulmonary function parameters: DLCO% (OR =0.693, 95% CI: 0.589–0.814, P<0.001) and FEV1% (OR =0.691, 95% CI: 0.579–0.826, P<0.001); and QCT metrics: EXP HAA% (OR =1.232, 95% CI: 1.000–1.517, P=0.049), fSAD ratio (OR =0.796, 95% CI: 0.692–0.916, P=0.001), and No. vessels_12mm (OR =0.564, 95% CI: 0.475–0.671, P<0.001). The combined diagnostic model demonstrated excellent performance. The area under the receiver operating characteristic (ROC) curve (AUC) reached 0.963 (0.920–0.998), indicating high discriminatory accuracy. Furthermore, the model 237 achieved an overall accuracy of 0.893, with a high sensitivity of 0.941 and a specificity of 0.818. The Hosmer-Lemeshow test yielded a statistic of 14.708 with a P value of 0.065, indicating good calibration and a reliable goodness-of-fit without significant overfitting (Figure 5 and Table 3). Performance and visualization of the Ridge regression model for CTD-ILD prediction. (A) ROC curve demonstrating the discriminative performance of the prediction model (AUC =0.963). (B) DCA assessing the net benefit of the model across different threshold probabilities compared with the “treat all” (red line) and “treat none” (green line) strategies. (C) Forest plot displaying the multivariable Ridge regression results for significant predictors. Blue circles indicate protective factors (OR <1) and red circles indicate risk factors (OR >1). Point sizes are proportional to the statistical significance [−log10(P)]. Error bars represent 95% confidence intervals. The dashed vertical line indicates the null effect (OR =1). (D) Bootstrap distribution of AUC values based on 1,000 resamples, with the red dashed line indicating the original AUC (0.963) and green dotted lines indicating the 95% CI (0.920, 0.998). AUC, area under the curve; CI, confidence interval; CTD-ILD, connective tissue disease-associated interstitial lung disease; DLCO, diffusing capacity of the lungs for carbon monoxide; FEV1, forced expiratory volume in the first second; fSAD, functional small airway disease; HAA, high-attenuation area; OR, odds ratio; ROC, receiver operating characteristic. Table 3Ridge regression analysis results for CTD-ILD prediction factorsVariableβS.EZPOR (95% CI)Intercept0.6140.2192.80.0051.847 (1.202–2.839)DLCO%−0.3670.082−4.451<0.0010.693 (0.589–0.814)FVC %−0.1460.105−1.3940.160.864 (0.704–1.061)FEV1%−0.3690.091−4.078<0.0010.691 (0.579–0.826)Age0.4230.1173.62<0.0011.527 (1.214–1.920)Sex0.2220.0892.4990.011.248 (1.049–1.485)IN LAA%0.1660.111.5040.131.180 (0.951–1.465)EXP LAA%−0.0670.105−0.6390.520.935 (0.760–1.149)EXP HAA%0.2080.1061.9640.0491.232 (1.000–1.517)fSAD ratio−0.2280.071−3.1950.0010.796 (0.692–0.916)No. vessels-12mm−0.5720.088−6.48<0.0010.564 (0.475–0.671)CI, confidence interval; CTD-ILD, connective tissue disease-associated interstitial lung disease; DLCO, diffusing capacity of the lungs for carbon monoxide; FEV1, forced expiratory volume in the first second; fSAD, functional small airway disease; FVC, forced vital capacity; HAA, high-attenuation area; LAA, low-attenuation area; No. vessels, the number of small pulmonary vessels; OR, odds ratio; S.E, standard error. Ridge regression analysis results for CTD-ILD prediction factors CI, confidence interval; CTD-ILD, connective tissue disease-associated interstitial lung disease; DLCO, diffusing capacity of the lungs for carbon monoxide; FEV1, forced expiratory volume in the first second; fSAD, functional small airway disease; FVC, forced vital capacity; HAA, high-attenuation area; LAA, low-attenuation area; No. vessels, the number of small pulmonary vessels; OR, odds ratio; S.E, standard error. Discussion This study, based on paired inspiratory-expiratory CT scans and utilizing QCT parameters of lung density, functional small airways, and the pulmonary vascular system, systematically reveals for the first time the multidimensional imaging features of CTD-ILD at the parenchymal, airway, and vascular levels. Through multivariate logistic regression analysis, the study further identifies FEV1%, HAA%-IN, HAA%-EXP, and the number of small vessels at depths of 6, 12, and 24 mm from the pleural surface as independent influencing factors for CTD-ILD, providing new imaging-based evidence for the identification and intervention of the disease. For the results of lung microstructure analysis showed that CTD with ILD had higher HAA%-IN and HAA%-EXP compared with the CTD without ILD (all P<0.001). The reason of this finding may be because that inflammatory cells infiltrate in the lung interstitium and alveolar space in the pathogenesis of CTD-ILD, leading to damage of alveolar epithelial cells, which may appear as ground glass opacity in the early stage of imaging. With the proliferation and activation of fibroblasts and myofibroblasts, excessive deposition of extracellular matrix is caused, and the loss is further aggravated, resulting in fibrosis (17). Moreover, the HAA represented both ground glass opacity and fibrosis, these areas may represent both irreversible fibrotic regions and reversible inflammatory region (18). Therefore, the HAA% of CTD-ILD measured on the inspiration and expiration CT images were both increased. Scharm et al. described increased HAA regions paired phase CT were also observed in patients with idiopathic pulmonary fibrosis (IPF), additionally, it was found that the ventilation and perfusion functions in HAA regions were reduced compared to normal-density regions (19). Small airways are defined as bronchioles with an internal diameter <2 mm, typically corresponding to airway generations beyond the 8th level. Pathological studies have shown bronchiolitis in patients with ILD (20). Although PFTs are commonly used to assess small airway involvement, they are limited by dependence on patient cooperation and low sensitivity—often detecting abnormalities only in advanced disease. CT thus offers a valuable alternative. Previous studies have commonly employed dual-phase CT imaging, combined with air trapping assessment and parametric response mapping (PRM), to quantify small airway dysfunction in patients with COPD and asthma (21,22). Compared with healthy individuals, these patients typically exhibit significantly elevated volumes of fSAD and fAT (23-25). Given that patients with CTD-ILD may also experience small airway dysfunction (26), one might expect their CT-derived quantitative metrics to be correspondingly increased. However, contrary to this expectation, our study found that CTD-ILD patients exhibited significantly lower LAA%-EXP, fAT volume, fSAD volume, and fSAD volume ratio compared to controls. This seemingly paradoxical finding likely stems from the density-threshold-based analytical method used in the current study. Specifically, fSAD regions were defined as voxels with attenuation greater than −950 HU on inspiratory CT and less than −856 HU on expiratory CT, while LAA%-EXP was defined as voxels below −856 HU on expiratory CT. In CTD-ILD patients, however, common pathological features such as ground-glass opacities and pulmonary fibrosis increase overall lung density, thereby altering the distribution of lung attenuation values (19). As a result, fewer voxels meet the predefined low-attenuation thresholds, leading to an artificial underestimation of LAA and fSAD measurements. These findings highlight that underlying pulmonary parenchymal abnormalities can substantially interfere with density-based QCT metrics. This limitation warrants careful consideration when interpreting CT-based small airway assessments, particularly in patients with concomitant ILD. Furthermore, while severe COPD-related emphysema was strictly excluded, isolated subpleural paraseptal emphysema can occasionally coexist with ILD. Theoretically, concurrent emphysema should increase the LAA; however, our finding of significantly reduced LAA%-EXP in the CTD-ILD cohort powerfully demonstrates that the extensive high-density fibrotic alterations completely overwhelm and mask the low-density contributions of any localized emphysematous changes. Moreover, the interpretation of these density-based metrics must be contextualized against the physiological impact of normal aging. Established literature demonstrates that normal lung aging is associated with a loss of elastic recoil, which typically leads to increased air trapping and higher EXP-LAA (27). Because our CTD-ILD cohort was significantly older, the confounding effect of normal aging alone would theoretically result in higher fSAD and LAA in the ILD group. However, our results demonstrated the exact opposite—a significant reduction in these metrics in the older ILD cohort. This paradox elegantly reinforces our hypothesis: the pathological fibrotic density shift driven by CTD-ILD possesses an overwhelming effect that completely reverses the expected radiological trends of normal aging. Pathological changes in pulmonary small vessels of CTD-ILD are characterized by vasculitis, vascular wall thickening, and fibrosis (28,29). Endothelial cell injury triggers inflammatory cell infiltration, leading to vascular remodeling, occlusion and loss of pulmonary small vessels. Immune complex deposition and autoantibody-mediated immune responses resulted in perivascular inflammation and fibrosis (30). These pathological alterations impair pulmonary microcirculation and exacerbate interstitial fibrosis, contributing to progressive lung function decline. Quantitative parameters of the vascular number and volume were decreased in CTD-ILD than in CTD without ILD, as determined in this study. We observed that the No. vessels, No. vessels CSA< 5 mm2 at 6, 12 and 24 mm depth from the pleural surface, BV1, BV5, BV10 and TBV were significant lower in the CTD with ILD compared with those in the CTD without ILD (all P<0.001). These changes in vascular density may precisely account for the significantly reduced DLCO observed in patients with ILD. Recent evidence confirms that progressive fibrotic destruction in interstitial pneumonia leads to quantitative pruning of these exact vessels, subsequently driving increased vascular resistance, and subsequently progress to PH (31,32). Consistent with prior IPF research (33), we observed that fibrosis severity was inversely correlated with vascular density, with reductions in subpleural small vessels corresponding to the extent of peripheral lesions. Thus, the observed reductions in small vessel counts and volumes directly reflect fibrotic obliteration of the pulmonary microvasculature rather than hemodynamic alterations. Notably, unlike previous studies, our work specifically focuses on the impact of CTD at the level of the pulmonary microvasculature. We quantified vascular volume across different vessel diameters and assessed vessel counts according to their spatial distribution within the lung parenchyma (34). Our study, consistent with prior IPF research, found that fibrosis severity was inversely correlated with vascular density, with a reduction in small vessels near the pleura corresponding to the extent of subpleural lesions Furthermore, by utilizing a robust penalized machine learning approach to rigorously eliminate multicollinearity, we identified age, sex, DLCO%, FEV1%, EXP HAA%, fSAD ratio, and No. vessels_12mm as significant independent diagnostic indicators for CTD-ILD. These independent predictors collectively reflect the multidimensional pathological landscape of CTD-ILD, encompassing pulmonary microvascular rarefaction, small airway dysfunction, and parenchymal fibrosis. The reduction in subpleural vessel counts and gas transfer capacity (DLCO%) suggests that microvascular destruction and compromised alveolar-capillary membrane integrity are central to disease pathogenesis, while the elevated EXP HAA% captures the fibrotic burden that characterizes established ILD. Importantly, the interplay between small airway disease (reflected by FEV1% decline and fSAD parameters) and progressive vascular pruning may create a vicious cycle of impaired microcirculation and deteriorating lung function, driving the onset and progression of interstitial fibrosis. While these findings enhance our mechanistic understanding of CTD-ILD and identify potential quantitative imaging and physiological biomarkers, their clinical utility for early detection or risk stratification requires prospective validation, particularly in CTD patients without overt ILD at baseline to establish their predictive value prior to clinical manifestation. We acknowledge that our investigation has several limitations. First, as a single-center, cross-sectional study with a relatively small sample size, our findings only demonstrate the diagnostic value of QCT metrics in differentiating existing ILD at the time of imaging. They do not establish a predictive capability for the future onset or progression of ILD. Further longitudinal studies are required to determine whether these baseline imaging biomarkers can predict incident ILD or treatment response over time. Second, significant baseline imbalances existed between the two groups regarding age, disease duration, and CTD subtypes. The CTD-ILD group was significantly older but had a shorter disease duration since diagnosis. This is largely driven by the stark subtype disparity: the non-ILD group was dominated by SLE, which typically has an earlier onset and longer survival, whereas the ILD group was predominantly composed of SSc patients, who often develop ILD rapidly. While this reflects real-world clinical demographics, this extreme subtype bias acts as a major confounder for QCT metrics in future studies, we aim to expand our cohort to include more cases, allowing us to further investigate whether there are differences in vascular and small airway pathology across different CTD sub-types. Conclusions QCT multidimensionally reveals the complex pathological alterations of parenchymal fibrosis, microvascular pruning, and small airway dysfunction in CTD-ILD. Our study demonstrates that integrating core QCT metrics with clinical and physiological parameters can construct a highly accurate early diagnostic model. Ultimately, these multidimensional, objective QCT parameters not only deepen our understanding of the simultaneous vascular and fibrotic damage mechanisms in early CTD-ILD, but also hold great promise as key imaging biomarkers for early identification, objective monitoring, and clinical stratification. Supplementary The article’s supplementary files as Acknowledgments We thank the colleagues in our department (Department of Radiology, The Second Affiliated Hospital of Naval Medical University) for their help in our study. Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the Institutional Review Board of The Second Affiliated Hospital of Naval Medical University (No. 2018L028) and individual consent for this retrospective analysis was waived. Footnotes Reporting Checklist: The authors have completed the TRIPOD reporting checklist. Available at https://jtd.amegroups.com/article/view/10.21037/jtd-2025-1-2705/rc Funding: This study was supported by Youth Medical Talents – Medical Imaging Practitioner Program, the Project of Shanghai Health Commission (No. 202340067); National Key Research and Development Program of China (No. 2022YFC2410000 & No. 2022YFC2410002); the National Natural Science Foundation of China (No. 82271994); the Military Commission Health Care Special Project (No. 22BJZ07); the National Health Commission Capacity Building and Continuing Education Center (No. YXFSC2022JJSJ010); the Shanghai Hospital Development Center (No. SHDC22022310-B); Navy Medical University Teaching Achievement Cultivation Project (No. JPY2022B15); Shanghai Changzheng Hospital Teaching Achievement Cultivation Project (No. JXPY2021B10); and Shanghai Leading Talent Program of Eastern Talent Plan (No. LJ2023094). Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://jtd.amegroups.com/article/view/10.21037/jtd-2025-1-2705/coif). The authors have no conflicts of interest to declare. Data Sharing Statement Available at https://jtd.amegroups.com/article/view/10.21037/jtd-2025-1-2705/dss References Available at https://jtd.amegroups.com/article/view/10.21037/jtd-2025-1-2705/prf Supplementary Materials The article’s supplementary files as Data Availability Statement Available at https://jtd.amegroups.com/article/view/10.21037/jtd-2025-1-2705/dss