C66-04 Quantitative Analysis of Computed Tomography Chest Predicts Presence of Pulmonary Hypertension in a Cohort of Idiopathic Pulmonary Fibrosis Patients

Abstract Rationale Pre-capillary pulmonary hypertension (PH) is common in patients with idiopathic pulmonary fibrosis (IPF). It is unknown whether the patterns of pulmonary vascular involvement in IPF differ and how this relates to the severity of PH and lung parenchymal involvement. Vascular pruning refers to the loss of small pulmonary blood vessels, a hallmark of PH. This results in a measurable reduction in the volume of vessels less than 5 mm (BV5) and less than 10mm (BV10). The aim of our study was to use quantitative CT imaging metrics to extract these pulmonary vascular measurements in patients with IPF to predict the presence of PH. Methods This was a retrospective cohort study of patients with IPF, both with and without PH. Pulmonary hypertension was defined as mPAP 20 mmHg on RHC. For all patients, RHC, pulmonary function tests, echocardiogram, and CT scan data were collected. Image analysis was performed using Coreline on CT images obtained at the time of the RHC. Metrics obtained via Coreline were blood vessels less than 5mm2 over blood vessels greater than 10mm2 (BV5/BV10 ratio) and fibrosis percentage. Clinical characteristics and quantitative CT metrics were compared between those with PH and those without PH via t-test. Associations and predictors of PH were identified with receiver operator characteristic (ROC) curve and multivariable logistic regression analyses. Results In our cohort of IPF patients, those who had PH compared to those who did not have PH based on RHC results had statistically significant lower BV5/B10 ratio on CT measurements and a higher percentage of fibrosis (p 0.05). ROC analysis demonstrated that BV5/BV10 (AUC 0.75, p 0.01) and fibrosis percentage (AUC 0.73, p 0.01) effectively predict PH (Figure 1). The optimal cutoffs as determined by Youden’s index were 0.50 for BV5/BV10 and 19.5% for fibrosis percentage. Multivariable logistic regression demonstrated that both BV5/BV10 0.50 (OR 17.3, 95% CI 2.07-44.9, p = 0.008) and fibrosis 19.5% (OR 5.3, 95% CI 1.61-17.44, p = 0.01) were independently and significantly associated with an increased risk of PH among IPF. Conclusions Quantitative CT metrics analyzing BV5/BV10 ratio and percentage of fibrosis both independently and significantly predict the presence of PH in our IPF cohort. These findings suggest that the development of PH is secondary to multiple mechanisms among patients with IPF, including both parenchymal involvement and vascular pruning indicative of pulmonary arteriopathy. This abstract is funded by: None