Liver Tumor Motion and Reconstructed Doses in a Clinical Trial of Respiratory-Gated Proton Therapy for Hepatocellular Carcinoma

Purpose: Pencil beam scanning proton therapy of hepatocellular carcinoma (HCC) may reduce the risk of radiation-induced liver disease compared to photon radiation therapy, but tumor motion and anatomic changes can distort the dose. This was investigated through daily reconstruction of the delivered dose in a clinical trial of gated HCC proton therapy. Methods and Materials: Sixteen patients with HCC included in a prospective phase II trial received 58 to 67.5 Gy (RBE) of proton therapy in 15 fractions with treatment planning in the exhale phase of a 4-dimensional computed tomography (4DCT) and exhale-gated treatment guided by an external gating block. Cone beam CT (CBCT)-guided setup was based on 2-3 implanted fiducial markers. At 3-8 fractions per patient, the 3-dimensional tumor motion during pretreatment and posttreatment CBCT acquisition was measured via the positions of the fiducial markers as a surrogate for the moving tumor. The tumor motion during treatment was estimated from the gating block motion using an external-internal motion correlation model constructed from the CBCTs. The delivered clinical target volume (CTV) dose was estimated by modeling tumor motion as proton spot position shifts and energy shifts in the treatment planning system. Effects of interfractional anatomic changes were assessed by dose recalculation on weekly control 4DCTs. Results: The mean (±SD) root-mean-square tumor position error during spot delivery was 1.1 ± 0.8 mm (left-right), 2.9 ± 1.2 mm (cranio-caudal), and 1.6 ± 0.7 mm (anterior-posterior). Motion increased the CTV homogeneity index, defined as 100% × (D2% – D98%)/D50%, by 4.1 ± 2.6 percentage points for individual fractions and 1.5 ± 2.1 percentage points when accumulated over all analyzed fractions. Control 4DCTs demonstrated minimal CTV dose variations. Conclusions: Intrafractional motion affected single-fraction tumor doses because of interplay effects that tended to smear out after more than 3 fractions. The tumor dose was robust to interfractional anatomic changes.