Serial automated quantitative CT analysis in idiopathic pulmonary fibrosis: functional correlations and comparison with changes in visual CT scores

To determine whether computer-based CT quantitation of change can improve on visual change quantification of parenchymal features in IPF. Sixty-six IPF patients with serial CT imaging (6-24 months apart) had CT features scored visually and with a computer software tool: ground glass opacity, reticulation and honeycombing (all three variables summed as interstitial lung disease extent [ILD]) and emphysema. Pulmonary vessel volume (PVV) was estimated by computer only. Relationships between changes in CT features and forced vital capacity (FVC) were examined using univariate and multivariate linear regression analyses. On univariate analysis, changes in computer variables demonstrated stronger linkages to FVC change than changes in visual scores (CALIPER ILD:R-2=0.53, p < 0.0001; Visual ILD:R-2=0.16, p=0.001). PVV increase correlated most strongly with relative FVC change (R-2=0.57). When PVV constituents (vessel size and location) were examined, an increase in middle zone vessels linked most strongly to FVC decline (R-2=0.57) and was independent of baseline disease severity (characterised by CT fibrosis extent, FVC, or DLco). An increase in PVV, specifically an increase in middle zone lung vessels, was the strongest CT determinant of FVC decline in IPF and was independent of baseline disease severity.