Longitudinal prediction of outcome in idiopathic pulmonary fibrosis using automated CT analysis

Abstract

AIMSTo evaluate computer-derived (CALIPER) CT variables against FVC change as potential drug trials endpoints in IPF.\n\nMETHODS71 Royal Brompton Hospital (discovery cohort) and 23 Mayo Clinic Rochester and 24 St Antonius Hospital Nieuwegein IPF patients (validation cohort) were analysed. Patients had two CTs performed 5-30 months apart, concurrent FVC measurements and were not exposed to antifibrotics (to avoid confounding of mortality relationships from antifibrotic use). Cox regression analyses (adjusted for patient age and gender) evaluated outcome for annualized FVC and CALIPER vessel-related structures (VRS) change and examined the added prognostic value of thresholded VRS changes beyond standard FVC change thresholds.\n\nRESULTSChange in VRS was a stronger outcome predictor than FVC decline when examined as continuous variables, in discovery and validation cohorts. When FVC decline ([≥]10%) and VRS thresholds were examined together, the majority of VRS change thresholds independently predicted outcome, with no decrease in model fit. When analysed as co-endpoints, a VRS threshold of [≥]0{middle dot}40 identified 30% more patients reaching an endpoint than a [≥]10% FVC decline threshold alone.\n\nCONCLUSIONSChange in VRS is a strong predictor of outcome in IPF and can increase power in future drug trials when used as a co-endpoint alongside FVC change.\n\nEthics committee approvalApproval for this study of clinically indicated CT and pulmonary function data was obtained from Liverpool Research Ethics Committee (Reference: 14/NW/0028) and the Institutional Ethics Committee of the Royal Brompton Hospital, Mayo Clinic Rochester and St. Antonius Hospital, Nieuwegein. Informed patient consent was not required.\n\nTake home messageChange in the vessel-related structures, a computer-derived CT variable, is a strong predictor of outcome in idiopathic pulmonary fibrosis and can increase power in future drug trials when used as a co-endpoint alongside forced vital capacity change.