Purpose: Dosimetric uncertainty in the surface and superficial regions is still a major concern for radiation therapy and becomes more important when using the inverse planning algorithm for IMRT. The purpose of this study was to measure dose distributions and to evaluate the calculation accuracy in the superficial region for different planning target volume (PTV) shrinkage methods for head and neck IMRT plans.
Methods: A spherical polystyrene phantom 160 mm in diameter (ball phantom) was used to simulate the shape of the head. Strips of superflab bolus with thicknesses of 3.5 and 7.0 mm were spread on the surface of the ball phantom. Three sets of CT images were acquired for the ball phantom without and with the bolus. The hypothetical clinical target volume (CTV) and critical structures (spinal cord and parotid glands) were outlined on each set of CT images. The PTVs were initially created by expanding an isotropic 3 mm margin from the CTV and then margins of 0, 3, and 5 mm were shrunk from the phantom surface for dosimetric analysis. Seven-field IMRT plans with a prescribed dose of 180 cGy and same dose constraints were designed using an Eclipse treatment planning system. Superficial doses at depths of 0, 3.5, and 7.0 mm and at seven beam axis positions (gantry angles of 0 degrees, 30 degrees, 60 degrees, 80 degrees, 330 degrees, 300 degrees, and 280 degrees) were measured for each PTV shrinkage margin using 0.1 mm ultrathin thermoluminescent dosimeters. For each plan, the measured doses were compared to the calculated doses.
Results: The PTV without shrinkage had the highest intensity and the steepest dose gradient in the superficial region. The mean measured doses for different positions at depths of 0, 3.5, and 7.0 mm were 106 +/- 18, 185 +/- 16, and 188 +/- 12 cGy, respectively. For a PTV with 3 mm shrinkage, the mean measured doses were 94 +/- 13, 183 +/- 8, and 191 +/- 8 cGy. For a PTV with 5 mm shrinkage, the mean measured doses were 86 +/- 11, 173 +/- 8, and 187 +/- 5 cGy. The comparisons indicated that more than 73.3% of the calculated points are with doses lower than the measured points and the difference of the dose becomes more significant in the shallower region. At 7.0 mm depth, the average difference between calculations and measurements was 2.5% (maximum 5.5%).
Conclusions: Application of the PTV shrinkage method should take into account the calculation inaccuracy, tumor coverage, and possible skin reaction. When the tumor does not invade the superficial region, an adequate shrinkage margin from the surface is helpful for reducing the skin reaction. As the tumor invades the superficial region, adding a bolus is a method better than only contouring PTV with skin inclusion.