P177 Computed tomography diagnostic model for diagnosis of pulmonary hypertension. (15th November 2017)

• Record Type: Journal Article
• Authors: Swift, AJ; Chin, M; Currie, B; Elliot, CA; Charalampopolous, A; Rajaram, S; Wild, JM; Johns, C; Kiely, DG
• Is Part Of: Thorax. Volume 72(2017)Supplement 3
• Page Start: A178
• Page End: A178
• Publication Date: 2017-11-15
• Subjects: Chest -- Diseases -- Periodicals; Thorax; Chest -- Diseases; Periodicals; Periodicals; 617.54
• DOI: 10.1136/thoraxjnl-2017-210983.319
• Languages: English
• ISSNs: 0040-6376

Abstract

Abstract : Introduction: Pulmonary hypertension (PH) is severe cardiorespiratory condition associated with poor prognosis with diagnosis reliant on invasive right heart catheterization (RHC). Several measurements on computed tomography (CT) have been shown to have diagnostic value in PH, however few studies have attempted to identify the added value of combining CT metrics for the diagnosis of PH. The aim of this study is to develop a composite diagnostic CT model for patients with suspected PH. Methods: Patients with suspected PH who underwent CT and RHC were identified. Standard axial and reconstructed images were used to derive CT metrics of cardiac and pulmonary vasculature anatomy. A derivation and validation cohort were randomly constructed to derive and test a binary logistic regression model of PH. Receiver operating characteristic (ROC) analysis assessed the diagnostic value of the model and individual metrics. Results: 491 patients were identified (derivation cohort n=247 and validation n=244). Main pulmonary arterial (MPA) diameter, right ventricular outflow tract (RVOT) thickness, right ventricular muscle area and interventricular septal (IVS) angle variables correlated strongest to mean pulmonary arterial pressure, r=0.458 (p<0.001), r=0.441 (p<0.001), r=0.481 (p<0.001) and r=0.622 (p<0.001), respectively. The diagnostic regression model included RVOT, IVS angle, MPA diameter, LV size and the interlobar artery to bronchus ratio. The area under the curve from ROC analysis was 0.931 (p=<0.001) in the derivation cohort and a 0.938 (p=<0.001) value in the validation cohort, more accurate the individual CT metrics (p<0.05). A highly sensitive threshold of 0 units had a sensitivity of 95% and specificity of 50% and a highly specific threshold of 3.3 units had sensitivity of 69% and specificity of 100%. Conclusion: A multivariate diagnostic model derived from axial CT images is accurate in suspected PH. The identified highly sensitive and specific thresholds may help in both patient screening and in selection for referral to specialist centres.