Document resource
Background Interstitial lung disease (ILD) is a significant cause of morbidity and mortality in people with systemic autoimmune rheumatic disease (SARD). Effective screening, accurate prognosis and monitoring are essential components of patient management. High-resolution computed tomography (HRCT) plays a role in all these aspects but is currently reliant upon expert radiological interpretation. Quantitative CT (qCT) using deep-learning may provide a powerful adjunct tool. We aimed to assess this in patients with SARD-ILD.Methods We conducted a retrospective, observational cohort study across 3 centres in England. Patients required a multidisciplinary team diagnosis of SARD-ILD, ≥1 HRCT and ≥2 pulmonary function tests (PFTs) between 2015 and 2022. Data were censored on 30/01/2025. Demographics, PFTs, pathology and mortality data were obtained from electronic patient records. Anonymised HRCT scans were analysed using deep-learning (Qureight Ltd), to quantify airway, fibrosis, ground glass and vascular abnormalities. qCT and PFTs within 90 days of each other were paired for baseline and serial Pearsons correlation assessment.For patients with ≥12 months follow-up Cox proportional hazards modelling identified baseline factors (age, gender, disease sub-type, ethnicity, PFTs and qCT) associated with progression free survival (PFS).Results 140 patients were included ( table 1). 533/580 (91.9%) scans had successful quantitative analysis. 378/533 and 359/533 had paired forced vital capacity (FVC) and transfer capacity (TLCO). FVC and TLCO correlated significantly with all qCT biomarkers even when adjusted for per patient differences (p<0.05). Strongest correlations were seen between fibrosis extent score (Fibr8) and FVC and TLCO (r=-0.52 and r=-0.56, p<0.05).When assessed longitudinally only vessel quantification (Vascul8) remained significant after adjustment for per patient differences (r=-0.42 and r=-0.52, p<0.05).On multivariate Cox proportional hazards analysis of 111/140 eligible patients (median follow-up 4.83 years, IQR 2.14–7.98) extent of ground glass abnormality (Glass8) at baseline was associated with reduced risk of PFS (HR 0.91; CI95% 0.84–0.98) and Fibr8 was associated with disease progression (HR 1.09; CI95% 1.03–1.15).Abstract S15 Table 1Demographics of 140 patients with systemic autoimmune rheumatic disease - interstitial lung disease MDT Diagnosis Number of Patients Age (mean ± SD) Male Ever smoker Median CCI FVC% predicted (± SD) TLCO% predicted (± SD) Polymyositis and systemic sclerosis overlap (Anti-PMSCL) 19 62.7 (9.5) 7 (37%) 8 (42.1%) 4 (1) 80.1 (20.2) 69.3 (25.2) Anti-synthetase syndrome 46 56.7 (13.2) 21 (46%) 19 (41.3%) 3 (1) 72.4 (19.9) 61.4 (19.1) Rheumatoid-arthritis associated ILD 23 65.2 (8.6) 17 (74%) 14 (60.9%) 4 (1.5) 88.0 (21.4) 65.7 (14.9) Sjogrens Syndrome 1 52.6 (n/a) 0 1 (100%) 3 (0) 66.6 75.5 Systemic sclerosis 51 55.3 (14.7) 13 (26%) 21 (41.2%) 3 (3) 76.4 (17.2) 61.8 (22.8) Total 140 58.4 (13.2) 58 (41%) 63 (45%) 4 (2) 77.4 (19.7) 63.4 (20.7) Legend: CCI= Charlsons comorbidity, FVC= Forced vital capacity, ILD= Interstitial lung disease, MDT= Multi-disciplinary team, PMSCL= Polymyositis-Scleroderma, TLCO= Transfer capacity for carbon monoxideConclusions QCT correlates significantly with changes in pulmonary physiology and appears useful in predicting disease progression in SARD-ILD. An adjunct role for qCT in identification, prognostication and disease monitoring of ILD in patients with SARD warrants prospective evaluation.