CO:10:4 | Lung vasculature quantification on computed tomography predicts new onset of interstitial lung disease in systemic sclerosis

Maria Iacovantuono^1|2, Nicholas Landini^2|3, Lisa Jungblut⁴, Gesa Sauer², Rucsandra Dobrota², Maria Sinziana Muraru-Carbune², Muriel Elhai², Carmen-Marina Mihai², Mike Oliver Becker², Maria Sole Chimenti¹, Thomas Frauenfelder⁴, Anna-Maria Hoffmann-Vold⁵, Oliver Distler², Cosimo Bruni². | ¹Rheumatology, Allergology and Clinical Immunology, Department of Systems Medicine, Tor Vergata University, Rome, Italy; ²Department of Rheumatology, University Hospital Zurich, University of Zurich, Switzerland; ³Department of Radiological Sciences, Oncology and Pathology, Sapienza University, Policlinico Umberto I, Rome, Italy; ⁴Institute for Diagnostic and Interventional Radiology, University Hospital Zurich, University Zurich, Switzerland; ⁵Department of Rheumatology, Oslo University Hospital, Norway

Interstitial lung disease (ILD) is common in systemic sclerosis (SSc) and a major cause of death. While clinical risk factors for ILD onset at 1-year and long-term follow-up have been identified (1,2), risk stratification remains suboptimal. Pulmonary vascular volume (PVV) on HRCT is linked to both presence and severity of SSc-ILD (3). Here, we investigate the potential of radiomics in identifying patients at risk of developing ILD. We included patients from a single referral centre, with no ILD on baseline HRCT and at least one clinical and HRCT follow-up. Images 1) were assessed for presence/absence of ILD at all timepoints, 2) underwent lung texture analysis (LTA,Imbio), quantifying PVV, normal lung, lung pathologies (hyperlucency, ground-glass, reticular, honeycombing), as percentage of the lung volume, for whole lungs and upper, middle, lower zones. Univariable and multivariable prediction models were developed to identify risk factors for ILD onset, using Cox regression and generalizes estimating equation (GEE) analysis for ILD onset ever and at 1-year, the latter allowing for multiple observations of the same patient. Both models were adjusted for the established clinical risk factors (1,2). Among 248 SSc patients with no ILD at baseline HRCT, new onset of ILD occurred in 54 (22%) cases over a median follow-up of 39(24-72)months. Patients developing ILD were more frequently males, with diffuse skin involvement, anti-topoisomerase I antibodies and shorter disease duration (Table I). Additionally, patients with new onset ILD over the study period presented higher values of PVV across all lung zones at baseline HRCT (Table II). In multivariable Cox regression analysis, new ILD onset was independently predicted by PVV% from the whole lung [HR 1.054 (1.030-1.078)], as well as from each individual zone. When integrating the PVV% of each single zone with the clinic risk factors in the multivariable analysis, we observed the significant, independent impact of the lower zone PVV% [HR 1.217 (1.113–1.332)], in addition to the known risk associated with clinical features [HR 3.281 (2.145–5.021)] (Figure I-A). Focusing on 1-year ILD onset, we analysed 279 follow-up visits from 193 patients and observed 22 (11.4%) new onset ILD (Table I). On univariable GEE, we confirmed the association of PVV from the whole lung on ILD onset, with particular focus on the middle and upper zones, alongside the clinical risk factors. In the multivariable model adjusted for clinical risk factors, PVV% of the upper zone remained significantly associated with ILD onset [OR 4.127 (1.016–16.761)] (Figure I-B). PVV% from HRCT scans of SSc patients without ILD can predict new onset of ILD at both 1-year and long-term follow-up, supporting the use of radiomics in risk phenotyping.

References.

1. Petelytska L, Bonomi F, Cannistrà C, Fiorentini E, Peretti S, Torracchi S, et al. Different definitions of disease severity, progression and activity in systemic sclerosis: data from the EUSTAR database. Arthritis Rheumatol. 2023;75(Suppl 9):Abstract 1234.

2. Hoa S, et al. Arthritis Rheumatol. 2025.

3. Occhipinti M, Bosello S, Sisti LG, Cicchetti G, de Waure C, et al. Quantitative and semi-quantitative computed tomography analysis of interstitial lung disease in systemic sclerosis. PLoS One. 2019;14(11):e0224982. doi:10.1371/journal.pone.0224982.