Automated high-resolution CT analysis outperforms visual assessment in predicting interstitial lung disease progression in SSc

Objectives: Progressive interstitial lung disease (ILD) is the leading cause of mortality in SSc. Early recognition of progression is crucial to timely treatment, but reliable early markers are lacking. Automated quantitative analysis of chest high-resolution CT (HRCT) may provide an objective tool to detect early progression. This study aims to compare automated and visual assessment in identifying early SSc-ILD progression and evaluate their predictive value for subsequent functional decline. Methods: In this retrospective longitudinal study, 33 patients with SSc-ILD underwent HRCT and pulmonary function tests (PFTs) at baseline (T1) and after 1 year (T2; IQR 0.75–2.5 years). PFTs were repeated 12 months after T2 (T3; IQR 9–14.5 months, T3). Two radiologists performed blinded semi-quantitative visual scoring of fibrosis and ground-glass opacity (GGO), while automated analysis was conducted using Thoracic VCAR (GE Healthcare). Associations with PFTs and prediction of functional progression (Erice/INBUILD criteria) and treatment escalation were analysed. Results: Both visual and automated analyses detected fibrosis progression (visual: 1.5 → 1.6, P = 0.0005; software: 0.34% → 0.4%, P = 0.01). Only automated analyses identified increases in GGO (15.5% → 15.8%, P = 0.04) and reduced normal lung tissue (75.6% → 73%, P = 0.009). Software-derived parameters correlated strongly with PFT changes, whereas visual scores showed weaker correlations. Early (T1–T2) automatically-detected GGO increases predicted subsequent functional decline at T3 (P = 0.04), with changes >50 ml (8.2%) yielding 81% sensitivity and 77% specificity. Conclusion: Automated quantitative HRCT analysis, but not visual evaluation, predicts long-term functional progression in SSc-ILD.