Abstract Rationale In interstitial lung disease, ground glass opacities (GGO) and fibrosis co-exist in varying proportions. However, the individual influence of GGO and fibrosis on disease trajectory and lung function change remains unclear. This work aimed to study the association between GGO and fibrosis HRCT biomarkers with forced vital capacity (FVC) in a dataset of treatment-naive patients with idiopathic pulmonary fibrosis (IPF). Methods HRCT scans of 377 patients from the Prospective Observation of Fibrosis in the Lung Clinical Endpoints Study (PROFILE) were segmented into GGO and fibrosis using a proprietary 3D convolutional neural network with a U-net architecture. Patients were classified into fibrosis-rich and GGO-rich based on the percentage of GGO and fibrosis in their lungs. The association of the derived HRCT biomarkers and FVC at baseline and follow-up was investigated using Pearson correlation and linear regression. Six- and twelve-month change in imaging biomarkers was determined with a time window of ± 3 months. Results Of the 377 PROFILE patients with paired HRCT and FVC data, 277 were classified as fibrosis-rich and 100 as GGO-rich (Figure 1A). Fibrosis-rich patients had a lower mean FVC than GGO-rich patients (Figure 1A). Baseline FVC had a moderate correlation with fibrosis percentage in fibrosis-rich patients and a weak correlation with GGO percentage in GGO-rich patients (Figure 1B). Six-month GGO (%) change in 34 patients with follow-up scans showed statistically significant linear association with six-month FVC (L) change (r = -0.66, p-value 0.001, adjusted R2 = 0.42, Figure 1C). No association of twelve-month FVC change with twelve-month GGO change was found (r = -0.090, p-value = 0.582, adjusted R2 = -0.018). There was no association between six-month Fibrosis (%) change and six-month FVC (L) change (r = 0.26, p-value = 0.144, adjusted R2 = 0.036). However, by 12 months there was a linear relationship between FVC change of 40 patients and fibrosis change (r = -0.38, p-value = 0.014, adjusted R2 = 0.12, Figure 1D). Conclusions Within a cohort of patients with IPF we identified a subgroup of patients with significant GGO. Longitudinal analysis suggests that short-term FVC change maybe driven by GGO-change, whereas in longer follow-up periods lung function change is driven by worsening fibrosis. The ability to quantify GGO and fibrotic components independently could help us understand some of the variability in disease trajectory and may have implications for clinical trial enrolment and interpreting treatment efficacy. This abstract is funded by: Qureight Ltd
Kirov et al. (Fri,) studied this question.