ESTRO 2024 - Abstract Book

S1235

Clinical - Head & neck

ESTRO 2024

Hadassah Medical Center, Radiation Oncology, Jerusalem, Israel

Purpose/Objective:

Adaptive radiation therapy (ART) offers a dynamic approach to address structural and spatial changes that occur during radiotherapy (RT) for locally advanced Head and Neck (HN) cancers. These tumors often undergo significant variations in target volumes and organs-at-risk (OARs) during treatment, necessitating plan adaptation. The integration of daily ART with Cone-Beam CT (CBCT) imaging presents a solution to enhance the therapeutic ratio by accommodating inter-fractional changes.

Material/Methods:

This retrospective study included 22 patients with LAHN cancers. The patients selected were those deemed suitable for daily adaptive radiotherapy (ART) aided by artificial intelligence (AI) on a daily basis using cone-beam computed tomography (CBCT). The study aimed to assess the feasibility of AI-assisted ART and evaluate its impact on treatment outcomes and toxicity. Radiotherapy Planning and Delivery: Target volumes and organs at risk (OARs) were contoured, with reference intensity-modulated radiation therapy (IMRT) plans created and approved according to institutional guidelines. All cases and plans were reviewed by clinicians prepared for adaptive treatments. CBCT-Based Adaptive Radiotherapy: Patients were aligned on the treatment table, and daily CBCT scans were obtained. These scans were registered with the simulation CT scan and an AI-based process was employed for auto contouring and structure deformation. The original gross tumor volumes were preserved utilizing pre-treatment imaging. Two plans, one scheduled and one adaptive, were generated for each fraction, with the superior plan chosen for delivery. Calculation-based quality assurance was performed for adaptive plans, followed by radiation delivery. Treatment Modalities: Patients received either chemoradiotherapy or radiation therapy alone, with systemic treatment protocols in place. Weekly monitoring and toxicity assessments were conducted throughout the treatment course, followed by regular follow-up evaluations. Data Analysis: Data analysis involved the use of R statistical software, with descriptive statistics for summarizing patient characteristics. Linear mixed-effects models were used to address repeated measurements, with Wald tests employed to assess fixed effects. Model assumptions were validated through diagnostic tools.

Results:

A total of 770 radiotherapy plans were generated, with the adaptive plan chosen for 95.7% of the fractions. The average time for an adaptive treatment session was 20 minutes, with a range of 18 to 23 minutes.

Comparing adaptive and scheduled plans, significant improvements were observed in dosimetric parameters. The mean V95 increased in adaptive plans for various target volumes, including PTV70, PTV60, PTV56, and PTV54 by 1.2%, 7.2%, 6.0%, and 1.5%, respectively.