Lens dose in IMRT treatment of the head and neck

Abstract

Listen

Translate article

Purpose: Intensity modulated radiotherapy (IMRT) is used to treat lesions in the head and neck where conventional delivery techniques can not adequately treat the tumor. Dose to the lens is a concern in IMRT treatments because very often the planner chooses unconventional beam angles in the planning process, where beams either enter or exit through the eye and/or lens. Additionally, there is typically a factor of 3 increase in monitor units delivered in IMRT treatments and the dose due to MLC transmission can be appreciable. Direct in-vivo measurement of the lens dose during IMRT treatments is impossible and dose calculation accuracy using conventional algorithms is limited in superficial tissues like the lens. The objective of this work is to investigate and determine the dose to the lens during step and shoot IMRT treatments. Materials and Methods: At our institution, the Corvus pencil-beam based inverse treatment planning system (Nomos Corp, Sewickley, PA) was used to develop step and shoot IMRT treatment plans for tumor sites including the nasopharynx, maxillary and ethmoid sinus, and parotid. Primary tumor prescription doses ranged from 55-64 Gy and were treated on a Varian 2100C linac with 4 MV X-rays. To date, 8 patients treated with IMRT have been studied. The lens was contoured and used for optimization in 4 cases with the dose limit set at 2-10 Gy. Mean dose to the lens was obtained from the values reported by the inverse planning system. Results from the inverse planning system were also calculated using Monte Carlo simulation of the IMRT treatment starting from the actual leaf-sequence files and using the CT data used for the inverse treatment plan. Both the Corvus and Monte Carlo systems were commissioned for the 2100C 4 MV beam. The calculated lens dose was also compared to a measurement using n-type silicon diodes (Sun Nuclear Corp, Melbourne, FL) during a single fraction of the IMRT treatment. The diode was placed over the mid-pupil of both the left and right eyes and taken as an approximation of the single fraction dose to the lens. Results: The lens dose due to MLC transmission and internal scatter was generally 1-2% of the prescription dose. For those cases where the target was near the optic apparatus, the mean dose to the lens ranged from 7.2% to 66.2% of the prescription dose which depends in part on the dose limitations set by the planner. This corresponds to 4.8 Gy to 42.7 Gy over the complete course of 25 - 30 fractions. The average difference in mean dose to the target between the inverse planning system and Monte Carlo verification results was 0.9% ± 2.5% (±1 σ). For the mean dose to the lens predicted by the two systems, the results agreed to within 3.2% ± 8.4% of the prescription dose. Similarly, the lens dose as measured by the diodes was within 1.2% ± 1.8% of the Monte Carlo results except for two cases. Some diode measurements were as much as 50% the prescription dose but these did not agree with either the inverse planning system or Monte Carlo verification results. For one case in which the Monte Carlo and inverse planning system results differed by 19.5% of the prescription dose, the Monte Carlo verification results agreed with the diode measurements. Conclusion: The lens dose can amount to an appreciable fraction of the treatment dose in head and neck IMRT treatments. Treatment plans should be developed that take this into consideration. The dose delivered to the lens depends more strongly on the dose-volume limitations set during optimization than on the number of beams or individual beam orientations since the dose due to MLC transmission and internal scatter is small. The Corvus pencil-beam based inverse planning results agree with Monte Carlo verification results and therefore would appear to be reliable dose estimates during IMRT planning and optimization. Diode measurements can be a useful verification of the lens dose predicted by the treatment planning system but they can not be relied upon in all situations. Future work is planned to develop a Monte Carlo based diode calibration to reduce this uncertainty.