3183 - Automated Lattice Radiation Therapy Planning for Head and Neck Cancer: A Novel SRS-Based Workflow and Dosimetric Comparison with VMAT

12:30pm - 01:45pm ET

Poster Hall - Exhibit Hall A

Screen: 25

Presenter(s)

Meiling Xu, - National Cancer Center/National Clinical Research Center for Cancer/Cancer Hospital & Shenzhen Hospital,Chinese Academy of Medical Sciences and Peking Union Medical College, Shenzhen, Guangdong

M. Xu¹, Y. Li^2,3, Y. Jia¹, E. Quan¹, W. Yan^4,5, J. Yang^2,6, and Z. Dai¹; ¹National Cancer Center/National Clinical Research Center for Cancer/Cancer Hospital & Shenzhen Hospital,Chinese Academy of Medical Sciences and Peking Union Medical College, Shenzhen, China, ²Foshan Chancheng Hospital, FoShan, China, ³Foshan Fosun Chancheng Hospital, Foshan, Guangdong, China, ⁴Baptisth health, Lexington, KY, ⁵College of Medicine, University of Kentucky, Lexington, KY, ⁶Junxin Oncology Institute, Foshan, Guangdong, China

Purpose/Objective(s): This study evaluates the feasibility and plan quality of an automated Lattice Radiation Therapy (LRT) planning workflow for head and neck cancers, utilizing a dedicated multi-target stereotactic radiosurgery (SRS) system, and compares its dosimetric performance against conventional volumetric modulated arc therapy (VMAT).

Materials/Methods: Three patients with oral cavity cancer were retrospectively selected. For each case, 8–12 spherical lattice vertices were automatically placed within the gross tumor volume (GTV) using an in-house developed close-packing software algorithm. Each vertex was prescribed 10 Gy per fraction for three total fractions. Two LRT plans were generated per patient: (1) an automated plan (Plan_auto) generated with Brainlab Elements Multiple Brain Mets SRS 4.5 on a Varian TrueBeam LINAC (6MV FFF) using 3–7 non-coplanar dynamic conformal arcs, and (2) a conventional VMAT plan (Plan_conv) created with Eclipse 13.6 utilizing 4–6 coplanar full arcs. Dosimetric comparisons included vertex coverage (V100%), peak-to-valley dose ratio (PVDR = D_2%/D_50% within the GTV), and doses to organs at risk (OARs) including the larynx, great vessels, esophagus, and skin. Monitor unit (MU) requirements were also compared.

Results: Automated LRT planning was successfully implemented for all cases. On average, Plan_auto provided superior and more consistent vertex coverage (V100%: 99.5% ± 0.1%) compared to VMAT (96.4% ± 2.1%), as well as a significantly higher peak-to-valley dose ratio (4.6 ± 0.7 vs. 3.8 ± 0.5). For OAR sparing, Plan_auto resulted in a lower mean dose to the larynx (1.3 ± 0.5 Gy vs. 2.8 ± 1.2 Gy) and a lower maximum dose to the skin (13.5 ± 2.1 Gy vs. 17.2 ± 3.5 Gy), along with reduced maximum dose to the great vessels (14.5 ± 1.8 Gy vs. 16.1 ± 2.3 Gy). Maximum dose to the esophagus was low for both techniques (1.09 ± 0.8 Gy for Plan_auto vs. 0.19 ± 0.1 Gy for Plan_conv). However, the non-coplanar Plan_auto required approximately twice the monitor units of Plan_conv (~13400 MU vs. ~6200 MU).

Conclusion: This study demonstrates the feasibility of an automated LRT planning workflow using a commercial multi-target SRS platform. The automated approach yielded plans with dosimetrically superior characteristics, including more robust vertex coverage, enhanced intra-tumoral dose heterogeneity (higher PVDR), and better overall OAR sparing compared to conventional VMAT. Critically, this automated approach promises to enhance clinical efficiency and promote plan standardization, providing a robust technical pathway for the broader implementation of LRT. These findings suggest that leveraging automated SRS tools is a viable pathway for high-quality LRT, though further investigation in a larger cohort is warranted.