Feasibility study of robotic hypofractionated lung radiotherapy by individualized internal target volume and XSight Spine Tracking: a preliminary dosimetric evaluation.

Abstract

To investigate the dosimetric impacts of lung tumor motion in robotic hypofractionated radiotherapy for lung cancers delivered through continuous tracking of the vertebrae by the XSight Spine Tracking (XST) mode of the CyberKnife. Four-dimensional computed tomography (4DCT) scans of a dynamic thorax phantom were acquired. Three motion patterns (one-dimensional and three-dimensional) of different range were investigated. Monte Carlo dose distributions were generated with 4DCT-derived internal target volume (ITV) with a treatment-specific setup margin for 12.6 Gy/3 fractions. Six-dimensional error correction was performed by kV stereoscopic imaging of the phantom's spine. Dosimetric effects of intrafractional tumor motion were assessed with Gafchromic films (Ashland Inc, Wayne, NJ, USA) according to 1) the percent measurement dose points having doses above the prescribed (P (> Dpres)), mean (P (> Dm)), and minimum (P (> Dmin)) ITV doses, and 2) the coefficient of variation (CV). All plans attained the prescription dose after three fractions despite marked temporal dose variations. The value of P (> Dpres) was 100% after three fractions for all plans, but could be smaller (~96%) for one fraction. The values of P (> Dmin) and P (> Dm) varied drastically interfractionally (25%-2%), and could be close to 0% after three fractions. The average CV ranged from 2.8% to 7.0%. Correlations with collimator size were significant for P (> Dmin) and P (> Dm) (P < 0.05) but not P (> Dpres) (P > 0.05). Treating lung tumors with CyberKnife through continuous tracking of the vertebrae should not be attempted without effective means to reduce the amplitude and variability of target motion because temporal dose variations owing to the intrafractional target motion can be significant.