Pulmonary artery filling defects: differentiating systemic-to-pulmonary artery shunt from pulmonary embolism.

BACKGROUND: Systemic-to-pulmonary artery shunt (SPS) may present pseudo pulmonary artery filling defects (PAFDs) on computed tomography pulmonary angiography (CTPA), which can mimic pulmonary embolism (PE) and thus potentially cause misdiagnosis and inappropriate treatment. METHODS: We screened patients who underwent CTPA between August 2020 and August 2024. Patients with PAFDs were included in the study and identified into either the PE or SPS group based on the delayed phases. The images and clinical characteristics of two groups were compared and the risk factors for SPS were evaluated. We further established an SPS Risk Scoring to predict SPS from PAFDs and validated the scoring in an independent cohort. Results Of 26,796 patients who underwent CTPA, 2,707 had PAFDs. SPS was identified in 273 patients (10.1%), of whom 265 (97.1%) were initially misdiagnosed with PE, and 44 (16.1%) received anticoagulation therapy. Independent risk factors for SPS included bronchiectasis, tuberculosis sequelae, pulmonary aspergilloma, hemoptysis, single PAFD, D-dimer < 2.505 mg/L, and the absence of lung cancer or deep vein thrombosis. The simplified SPS Risk Score showed excellent discrimination (AUC = 0.986), with an optimal cut-off of 3.5 points (sensitivity: 95.5% [253/265], specificity: 94.1% [2,186/2,324]), and scores > 3.5 favoring SPS. In the independent validation cohort, the score achieved a sensitivity of 100% [31/31], and a specificity of 89.7%[182/203]. Conclusion For patients with PAFDs and high-risk features identified by the SPS Risk Score, delayed-phase CTA should be performed to accurately differentiate SPS from PE and prevent inappropriate anticoagulation.