AI-derived cardiothoracic structures and coronary artery calcification volumes on non-contrast CT as predictors of mortality in pulmonary hypertension

Abstract Objective To assess the prognostic value of AI-derived cardiothoracic structures and coronary artery calcification volumetric measurements in pulmonary hypertension (PH) using non-gated, non-contrast CT. Materials and methods Patients with suspected PH who underwent non-gated, non-contrast chest CT and right heart catheterisation within one month were retrospectively identified from the Assessing the Spectrum of PH Identified at a Referral Centre (ASPIRE) registry. Two validated deep learning models were used to segment and measure the volumes of cardiothoracic structures and coronary artery calcifications. The prognostic value of AI-derived cardiothoracic structures and coronary artery calcification volumetric measurements was assessed using univariable and multivariable Cox regression adjusted for demographics and PH subgroups and REVEAL scores. Results Six hundred twenty patients were included (mean age 64 ± 13 years; 62% female; 388 patients died during follow-up), comprising 202 patients with pulmonary arterial hypertension, 84 with PH-LHD, 112 with PH associated with lung disease, 65 with chronic thromboembolic PH, 9 with unclear PH, and 148 patients without PH. Calcifications were present in the right coronary artery (RCA), mainstem and left anterior descending artery (LAD), and left circumflex artery (LCX) in 45%, 67%, and 35% of patients, respectively. In multivariable Cox regression analysis adjusted for age, sex, and PH subgroup, right atrial (RA) volume was independently associated with mortality (HR 1.024, 95% CI: 1.009–1.039, p = 0.001), as were coronary artery calcifications in the RCA (HR 1.378, 95% CI: 1.113–1.705, p = 0.003), mainstem and LAD (HR 1.868, 95% CI: 1.441–2.421, p < 0.001), and LCX (HR 1.342, 95% CI: 1.084–1.660, p = 0.007). After further adjustment for the REVEAL risk score, these associations remained significant (RA volume HR 1.026, RCA calcification HR 1.604, mainstem and LAD calcification HR 2.339, LCX calcification HR 1.576; all p < 0.001). Conclusion AI-derived RA volume and coronary artery calcifications are independently associated with increased mortality in PH patients. Key Points Question The prognostic significance of AI-derived cardiothoracic volumes and coronary artery calcification measured on routine non-gated, non-contrast chest CT in patients with PH remains uncertain. Findings AI-derived RA volume and coronary artery calcification volumes independently predicted mortality in PH patients, even after adjustment for demographics, PH subgroups, and REVEAL score. Relevance statement Automated CT-based AI measurements can provide additional prognostic biomarkers to improve risk stratification in PH beyond established clinical scores.