3576 - Funnel-Shaped PTV Margins for Optimized Target Coverage and OAR Sparing in Nasopharyngeal Carcinoma Radiotherapy

Purpose/Objective(s):

Conventional uniform PTV margins in NPC radiotherapy often fail to balance target coverage and OAR protection, leading to underdosing in the lower neck and overdosing of critical structures like the brainstem. This study developed a Python-automated funnel-shaped PTV margin (FPM) protocol. We compared its dosimetric performance to the conventional 3-mm uniform margin, aiming to provide an optimized approach for PTV delineation.

Materials/Methods:

This retrospective study enrolled 20 patients with locally advanced NPC who received radiotherapy at our institution in 2025. Two sets of PTVs were generated: one using the FPM protocol and the other using conventional 3-mm uniform margins. In the FPM protocol, the target volume was divided into superior and inferior segments at the mid-level of the odontoid process. The superior segment received uniform 1-mm margins in the left-right and anterior-posterior directions. The inferior segment was further subdivided into five subsegments along the craniocaudal axis, with margins progressively increasing from 2 mm to 6 mm at 1-mm intervals. All PTVs (PTVnx, PTVnd, PTV1, and PTV2) were generated via the developed FPM software. FPM plans (FP) and conventional PTV margin plans (CP) were subsequently created. Key dosimetric parameters (D2%, D98%, Dmean) for target volumes and radiation doses to critical OARs between the two planning approaches were compared. Throughout radiotherapy, all patients underwent daily online cone-beam computed tomography for setup verification and correction.

Results:

FP plans exhibited a steeper dose gradient within the target volumes compared to CP plans. In terms of target volume dosimetry, FP plans achieved significantly higher D98% (69.90 ± 0.24 Gy) and significantly lower D2% (74.77 ± 0.27 Gy) and Dmean (72.65 ± 0.22 Gy) in the PTVnx. For the PTVnd, FP plans demonstrated significantly reduced D2% (75.01 ± 0.39 Gy), D98% (69.46 ± 0.20 Gy), and Dmean (72.60 ± 0.25 Gy). In the PTV1, FP plans yielded significantly lower D2% (74.15 ± 0.45 Gy) and Dmean (66.00 ± 0.83 Gy). No statistically significant differences were observed between the two planning approaches for any dosimetric parameter in the PTV2.The radiation doses to all critical OARs were significantly lower in the FP plans compared to the CP plans (all P < 0.001) (Table1).

Conclusion:

The Python-automated FPM protocol ensures adequate target coverage while significantly reducing radiation doses to critical OARs. This dual optimization of tumor targeting and normal tissue sparing provides an efficient, feasible strategy for precise PTV delineation in NPC radiotherapy.

Table1. Dosimetric Comparison of Organs at Risk: Funnel-Shaped vs. Conventional PTV Margins

OARs	Parameter	FP (Gy)	CP (Gy)	P value
Brain Stem	Dmax	51.29±7.73	58.48±7.33	<0.001
	D1%	47.12±7.09	54.54±7.83	<0.001
Spinal Cord	Dmax	41.83±3.77	45.22±4.78	<0.001
Parotid_L	Dmean	39.63±5.18	43.07±4.85	<0.001
Parotid_R	Dmean	37.99±4.76	41.17±4.60	<0.001
Chiasm	Dmax	51.51±16.07	55.68±13.49	<0.001