Interstitial lung disease (ILD) comprises a heterogeneous group of parenchymal lung disorders with diverse clinical trajectories [1]. Idiopathic pulmonary fibrosis (IPF) represents the prototypical progressive fibrosing ILD, characterized by steady lung function decline and poor survival [2,3]. Although non-IPF ILDs, including connective tissue disease-associated ILD (CTD-ILD), often exhibit a more inflammation phenotype and better prognosis, a substantial subset develops progressive pulmonary fibrosis (PPF), sharing clinical behavior and outcomes with IPF [3–5]. Early identification of PPF in CTD-ILD remains challenging but clinically important, as progression is associated with increased morbidity and mortality.

In 2022, international guidelines from the ATS/ERS/JRS/ALAT defined PPF based on combination of worsening symptoms, physiologic decline, and radiologic progression [3]. Radiologic progression was defined as an increase extent or new fibrotic lesions. However, in clinical practice, assessment of radiologic progression relies largely on visual interpretation, which is subject to interobserver variability and lacks standardized quantitative thresholds or defined follow-up intervals. While automated quantitative CT analysis has shown prognostic value in fibrotic ILAs, including IPF, by providing objective measures of fibrotic burden and disease progression [6,7], quantitative indices derived from CT imaging, such as histogram-based or density-derived metrics, have also been associated with disease severity and mortality [7–9]. These approaches offer improved reproducibility and may overcome limitations of subjective visual assessment. However, validated thresholds for radiologic progression in CTD-ILD are not established. We therefore aimed to identify an objective quantitative CT threshold for radiologic progression in CTD-ILD and to evaluate whether integrating CT-derived fibrosis progression with pulmonary function parameters improves identification and prognostic stratification of PPF.

We conducted a retrospective cohort study of patients diagnosed with CTD-ILD patients and followed at a tertiary referral center between 2019 and 2024. All patients were evaluated in the pulmonology clinic and received a multidisciplinary diagnosis of CTD-ILD, integrating clinical, radiologic, and rheumatologic assessments. Patients were included if they had at least two serial pulmonary function tests (PFTs) and high-resolution CT (HRCT) scans performed at approximately one-year intervals and suitable for quantitative analysis. CTD diagnoses were established by rheumatologists according to standard classification criteria [10–12], and ILD diagnose were confirmed by multidisciplinary discussion (MDD) that included radiologists, rheumatologists, and pulmonologists. Immunomodulatory therapy was prescribed by rheumatologists as standard treatment for the underlying CTD and ILD progression was independently monitored by pulmonologists using serial chest CT and PFTs.

Demographic characteristics, comorbidities, medication history, and laboratory data were extracted from medical records. PFTs were performed according to ATS/ERS standards [13,14]. Overall survival was defined from ILD diagnosis to death from any cause.

Chest CT images were analyzed using a commercially available deep learning-based software (Aview, version 1.1.43.7929, Coreline Soft) that automatically classified lung parenchyma into normal lung, emphysema, ground-glass opacity (GGO), consolidation, reticular opacity, and honeycombing. Pattern percentages were calculated by measuring segmented volumes relative to total lung volume. This was achieved by multiplying the pixel count for each pattern by the slice thickness and subsequently dividing by the total voxel number of the entire lung (Supplementary Fig. 1) [15]. Fibrotic ILD extent was quantified using three definitions: (1) combined percentage of reticular opacity and honeycombing (Definition); (2) reticular opacity, honeycombing, and GGO (Definition 2); and (3) reticular opacity, honeycombing, GGO, and consolidation (Definition 3). Changes in quantitative ILD extent (QILD) were calculated over one-and two-year intervals.

PPF was defined according to guideline-based criteria [3], when at least two of the following three criteria occurred within one year: (a) worsening of respiratory symptoms; (b) ≥5% reduction in the absolute value of forced vital capacity (FVC) or ≥10% decrease in diffusing capacity for carbon monoxide (DLCO); (c) radiologic progression signs. Visual determination of radiologic progression is established if at least one of the following six indicators is observed: (a) increased extent or severity of traction bronchiectasis and bronchiolectasis, (b) new GGO with traction bronchiectasis, (c) appearance of new fine reticulation, (d) increased extent or coarseness of reticular abnormality, (e) new or progressed honeycombing, and (f) further lobar volume loss.

Receiver operating characteristic (ROC) analysis was used to identify thresholds for CT-defined radiologic progression. Survival analyses were performed using Kaplan–Meier analysis, and multivariable models adjusted for age, sex, smoking status, and radiologic usual interstitial pneumonia (UIP) pattern. All statistical analyses were conducted using STATA software (ver. 16; StataCorp, College Station, TX, USA).

A total of 91 CTD-ILD patients were included (median age 63 years; 82% female). Primary Sjögren's disease (pSS, 26.4%), systemic sclerosis (SSc, 23.1%), and rheumatoid arthritis (RA, 22.0%) were the most common underlying CTDs (Supplementary Table 1). Most patients exhibited a non-UIP pattern on CT, predominantly nonspecific interstitial pneumonia (NSIP). Nearly all patients received immunomodulating therapy, while antifibrotic therapy was infrequent. At baseline, QILD using definition 1, 2, and 3 was 2.42%, 5.34%, and 5.62%, respectively.

Over follow-up, 30 patients (32.9%) met physiologic criteria for PPF. Baseline demographics, CT patterns, and baseline QILD did not differ significantly between PPF and non-PPF groups, although DLCO was higher in the PPF group at baseline (Supplementary Table 1). At one year, only Definition 1 demonstrated significantly greater fibrotic progression in PPF patients compared with non-PPF patients (0.60% vs. −0.20%, p=0.001, Fig. 1). At two years, fibrotic progression was consistently greater in PPF patients across all three definitions, but Definition 1 most clearly discriminated PPF from non-PPF in both absolute extent and longitudinal change. Mortality was higher in PPF patients (4/30, 13.3%) compared to non-PPF patients (1/61, 1.6%).

Fig. 1.

CT-quantified fibrosis progression by PPF status by Definition 1. Box plots represent median and interquartile range. Patients with PPF showed significantly greater fibrosis progression at both 1 year (A) and 2 years (B).

ROC analysis identified a 2.20% increase in fibrotic extent by Definition 1 over two years as the optimal threshold for radiologic progression associated with physiologic PPF (Area under the ROC curve [AUC]=0.68, Supplementary Fig. 2). Internal validation using bootstrap resampling (1000 iterations) confirmed the stability of the optimal cutoff (95% confidence interval [CI], −0.46% to 2.92%), with a bootstrap-corrected AUC of 0.68 (95% CI, 0.56–0.80) (Supplementary Fig. 3). Using this threshold, CT-defined radiologic PPF was present in 40% of physiologic PPF patients compared with 8.2% of non-PPF patients. Patients meeting the CT-defined PPF threshold showed significantly greater annual declines in FVC (−163.34 vs. 25.54mL/year; p<0.001) and DLCO (−2.99 vs. 1.03%/year; p<0.001). While forced expiratory volume in one second (FEV1) decreased in both groups without significant difference, FEV1/FVC ratio remained unchanged. Total lung capacity (TLC) declined in radiologic PPF patients, while ratio of residual volume (RV)/TLC increased in both PPF and non-PPF group, suggesting possible small airway involvement (Supplementary Fig. 4).

When patients were stratified using combined CT and PFT criteria, those fulfilling both radiologic (≥2.2% QILD increase) and physiologic criteria exhibited more severe functional impairment at baseline and significantly worse survival (Fig. 2, Supplementary Table 2). In multivariable analysis, combined CT and PFT-defined PPF emerged as the only independent predictor of mortality (adjusted odds ratio [OR]=43.88; 95% CI, 2.45–786.39; p=0.010, Supplementary Table 3).

Fig. 2.

Kaplan–Meier survival estimates by PPF status defined by radiologic and/or physiologic criteria. CT, computed tomography; PFT, pulmonary function test; PPF, progressive pulmonary fibrosis.

The study demonstrates that a modest increase in radiologic fibrotic burden, specifically reticulation and honeycombing, is associated with physiologic decline and adverse outcomes in CTD-ILD. The identified threshold of a 2.2% increase in fibrotic extent over two years is lower than previously reported thresholds in CTD-ILD [9], likely reflecting differences in baseline fibrosis burden, composition of CTD, and widespread use of immunosuppressive therapy in our cohort. Importantly, our findings highlight the limitations of relying on either radiologic or physiologic criteria alone. A substantial proportion of patients showed discordant progression, underscoring the need for integrated assessment to avoid under-recognition of clinically meaningful disease progression. In CTD-ILD, progression may occur in the context of active systemic inflammation or independent fibrotic progression, and distinguishing between these processes can be challenging. This highlights the importance of multidisciplinary assessment integrating clinical, functional, and imaging findings.

The observation that GGO and consolidation contributed little to longitudinal progression supports the concept that inflammatory or potentially reversible components should be distinguished from irreversible fibrotic features when monitoring PPF. Among the evaluated definitions, fibrosis defined by reticulation and honeycombing provided the most consistent and clinically relevant signals for progression in CTD-ILD.

Decline in FVC and DLCO are established indicators of progression in fibrotic ILD [3,16,17]. In our cohort, PPF patients showed greater reductions in these parameters, along with a more pronounced decline in FEV1. The observed increase in FEV1/FVC likely reflects restrictive physiology due to declining FVC. In contrast, increase in RV/TLC ratios suggested the presence of air trapping, which may indicate small airway involvement, as previously described in autoimmune-related ILDs such as RA and pSS [18–20]. As baseline obstructive features were comparable between groups, these findings may reflect progressive airway involvement rather than pre-existing airflow limitation. However, this interpretation should be considered exploratory, as direct assessments of small airway function, such as impulse oscillometry or expiratory CT, were not available.

This study has several strengths. To our knowledge, it is the first to propose an integrated CT-PFT framework for defining PPF in CTD-ILD, but several limitations warrant consideration. The retrospective, single-center design limits generalizability, and the small number of mortality events raises the possibility of model overfitting. Treatment effects could not be fully assessed due to limited use of antifibrotic therapy, which was largely related to reimbursement restrictions and regulatory indications during the study period rather than clinical selection bias. Quantitative HRCT analysis relies on specialized software that is currently more commonly used in research settings and may not be widely available in routine clinical practice. Prospective multicenter studies are required to validate the proposed quantitative threshold and integrated CT-PFT framework, and to determine whether early identification of CT-defined PPF can guide timely therapeutic interventions.

In conclusion, approximately one-third of CTD-ILD patients developed PPF in this cohort. A 2.20% increase in CT-quantified fibrotic extent over two years provided an objective marker of radiologic progression and, when combined with lung function decline, identified patients at markedly increased risk of mortality. These findings support an integrated CT and PFT approach for early detection and risk stratification of PPF in CTD-ILD.