RATIONALE: There is an urgent need for reliable biomarkers to predict progression in Idiopathic Pulmonary Fibrosis (IPF). While quantitative CT metrics, automated CT phenotypes, and proteomic biomarkers have been reported, their combined prognostic utility remains unassessed. This study evaluates CT metrics and a proteomic biomarker panel in a prospective IPF cohort (PROFILE). METHODS: Participants with analysable CT scans were included. Fibrosis volume (Fibr8) as a percentage of total lung volume was quantified using a validated deep learning model. %UIP probability was evaluated using the deep learning classifier, SOFIA. Proteomic markers, with missing values <15%, were included: SPD, MMP7, PRO-C3, PRO-C4, PRO-C6, CYFRA211, and CEA. Univariate ROC-AUC curves were used to determine predictors of 12-month progression (defined as FVC decline ≥10% or death). Survival analysis was performed using Cox proportional hazards and Kaplan-Meier estimator. RESULTS: In PROFILE (n=145), on multivariable analysis, higher baseline levels of each biomarker were associated with increased mortality, adjusting for baseline ppFVC: CYFRA21-1 (HR 1.02; p<0.005), SPD (HR 1.01; p<0.005), ProC6 (HR 1.05; p=0.01), Fibr8 (HR 1.03, p=0.02), and SOFIA UIP probability (HR 2.58, P<0.005). Proteomics provided significant improvement in explaining variation of survival over imaging alone (likelihood ratio test p<0.005). Fibr8 and CYFRA21-1 had the strongest associations with 12-month progression among imaging and proteomic biomarkers. A composite risk score (CRS) was developed by fitting Fibr8 and CYFRA21-1 against 12-month progression on logistic regression: CRS=(Fibr8[asterisk]0.044) + (CYFRA21-1[asterisk]0.341). The CRS was more strongly associated with mortality and provided an improved fit over each variable in isolation when controlling for disease severity using ppFVC: CRS (HR 1.65; p<0.005; AIC 937.26), CYFRA21-1 (HR 1.21; p<0.005; AIC 947.04), Fibr8 (HR 1.04; p<0.005; AIC 942.03), ppFVC alone (HR 0.98; p<0.005; AIC 956.32). Thresholds for each biomarker, optimised against 12-month progression separated the cohort into 2 prognostically distinct groups, controlling for ppFVC: Fibr8 (n=383, 15.3%, AUC 0.69, HR 2.21), CYFRA21-1 (n=145, 2.95 ng/ml AUC 0.68, HR 1.56). The CRS improved patients' prognostic separation over Fibr8 or CYFRA21-1 thresholds in isolation CRS (n=145, 2.09 score, AUC 0.71, HR 2.86) (Figure 1). Patients separated by CRS threshold had different ppFVC decline (p=0.01, 6.79 vs 15.69 pp/yr below/above threshold).CONCLUSION: Automated quantification of CT phenotype, fibrosis extent and protein biomarkers of epithelial injury and collagen turnover (CYFRA21-1, SPD, and ProC6) predict mortality in IPF, independently of imaging biomarkers and ppFVC. A composite risk score, combining radiological and proteomic biomarkers can be generated to provide improved prognostication in IPF.