Large scale plasma proteomic profiles improve prediction of idiopathic pulmonary fibrosis in general population.

BACKGROUND: Idiopathic pulmonary fibrosis (IPF) is a progressive interstitial lung disease of unknown etiology, with poor prognosis and substantial global economic burden. Circulating proteomics holds promise for unraveling IPF pathogenesis and identifying therapeutic targets. Herein, we aimed to explore the relationship between plasma proteomics and IPF, and evaluate whether proteomics could improve IPF risk prediction. METHODS: This cohort study included 44,306 participants from the UK Biobank. The development cohort, consisting of 39,035 participants from England, was randomly divided into a 7:3 training-testing ratio. The validation cohort included 5,211 participants from Scotland and Wales. Multivariable-adjusted Cox regression models were used to explore associations between 2,920 plasma proteins and incident IPF. In the training set (27,366 participants; 295 IPF cases), an IPF protein risk score (PRS) was constructed incorporating 256 proteins selected using least absolute shrinkage and selection operator (LASSO) penalty. Predictive performance was assessed using Harrell's C-index, time-dependent area under the receiver operating characteristic curve, continuous/categorical net reclassification improvement, and integrated discrimination improvement. RESULTS: The median follow-up duration was 13.6 years. We observed 464 proteins associated with IPF risk, primarily involved in pathways related to external side of plasma membrane, leukocyte cell-cell adhesion, cytokine activity, and cytokine-cytokine receptor interaction. CCL21 and CXCL9 were identified as key proteins within the protein network. In the testing set (11,729 participants; 142 IPF cases), integrating 256 LASSO-selected proteins (C-index increase 0.207; 95% CI 0.188, 0.224) and a weighted IPF-PRS (C-index increase 0.105; 95% CI 0.085, 0.112) significantly enhanced IPF prediction compared to traditional risk factors alone (C-index, 0.779; 95% CI 0.742, 0.817). Adding LASSO-selected proteins had the largest C-index of 0.986 (95% CI 0.977, 0.995), and significantly improved the continuous 10-year net reclassification (0.462; 95% CI 0.359, 0.534) and 10-year integrated discrimination index (0.747; 95% CI 0.141, 0.898). These results were verified in the external validation cohort (5,211 participants; 61 IPF cases). CONCLUSIONS: Our study characterized early proteomic contributions to IPF, and demonstrated that plasma proteomic data significantly enhance IPF risk prediction beyond traditional risk factors.