CT-Based Intratumoral Artery-Like Volume for Pretreatment Lung Shunt Estimation in Hepatocellular Carcinoma

To determine the clinical utility of automated 3-dimensional (3D) computed tomography (CT)-based quantification of intratumoral artery-like volume (IAV) for the preprocedural assessment of transarterial radioembolization (TARE) eligibility by ruling out high lung shunt fraction (LSF) in patients with large hepatocellular carcinoma (HCC).

This single-center retrospective study included patients with a single-nodular HCC (≥5 cm) who underwent both multiphasic CT and LSF measurement (2009-2024). LSF values were derived from technetium-99m macroaggregated albumin (MAA) scintigraphy. Automated 3D segmentation on arterial-phase CT defined tumor and aorta; IAVs were quantified using 5 thresholds (65%-85% at 5% intervals) relative to mean aortic Hounsfield unit. IAV metrics were compared across LSF groups (≤10%, 10%-20%, and >20%). Receiver-operating characteristic (ROC) analysis evaluated performance for LSF of >10% and >20% and determined cutoffs at fixed 80% and 90% sensitivities.

Of 169 patients, 123, 26, and 20 had LSF of ≤10%, 10%-20%, and >20%. IAV metrics at all thresholds showed a significant stepwise increase across these groups (P < .001). For predicting LSF of >10% and >20%, the area under the ROCs of IAV metrics ranged from 0.732 to 0.741 and 0.790 to 0.806, respectively. Using IAV75% as a representative metric, fixed 80% and 90% sensitivities yielded specificities of 50.4% (62/123) and 39.0% (48/123) for LSF of >10% (cutoffs, 0.06 and 0.02 cm³) and 74.5% (111/149) and 57.7% (86/149) for LSF of >20% (cutoffs, 0.50 and 0.16 cm³), respectively.

Automated CT-based IAV enables objective LSF risk stratification in large HCC and can identify low-risk patients preprocedurally, supporting a more streamlined TARE workflow.