Variability of a peripheral dose among various linac geometries for second cancer risk assessment

Abstract

Second cancer risk assessment for radiotherapy is controversial due to the large uncertainties of the dose-response relationship. This could be improved by a better assessment of the peripheral doses to healthy organs in future epidemiological studies. In this framework, we developed a simple Monte Carlo (MC) model of the Siemens Primus 6 MV linac for both open and wedged fields that we then validated with dose profiles measured in a water tank up to 30 cm from the central axis. The differences between the measured and calculated doses were comparable to other more complex MC models and never exceeded 50%. We then compared our simple MC model with the peripheral dose profiles of five different linacs with different collimation systems. We found that the peripheral dose between two linacs could differ up to a factor of 9 for small fields (5 × 5 cm2) and up to a factor of 10 for wedged fields. Considering that an uncertainty of 50% in dose estimation could be acceptable in the context of risk assessment, the MC model can be used as a generic model for large open fields (≥10 × 10 cm2) only. The uncertainties in peripheral doses should be considered in future epidemiological studies when designing the width of the dose bins to stratify the risk as a function of the dose.

A review of uncertainties in radiotherapy dose reconstruction and their impacts on dose–response relationships

“…Beyond 3.75 cm from the field edge, discrepancies were about 40% and could be up to 55% for the furthest points (11.25 cm in this study). In a similar comparison, Joosten et al [44] found differences of up to 179% for points within 10 cm from the beam edge of a small field size (5 cm×5 cm) when comparing the dose calculated with the TPS CMS XiO (version 4.60, Elekta, Crawley, UK), using a point kernel-based algorithm, and the dose measured with a 0.13 cm 3 ionisation chamber (IC10, Scanditronix Wellhofer GmbH, Schwarzenbruck, Germany).…”

“…The main algorithms available in present clinical TPSs [55] are measurement-based [56,57] and model-based algorithms [58][59][60][61][62][63][64]. However, TPSs are not designed to evaluate the dose in the regions of the patient's body outside the treatment beams [44,[65][66][67][68][69][70][71][72][73][74][75]. This has led medical physicists to develop in-house algorithms for estimating out-of-field doses.…”

“…Regarding out-of-field dose evaluations from TPSs, many authors have reported that TPSs should not be used for out-of-field dose evaluations [44,[65][66][67][68][69][70][71][72][73][74][75]. Howell et al [67] quantified the discrepancies between TLD measurements and dose calculations using Eclipse TPS (version 8.6, Varian Medical Systems) and the AAA algorithm, a point kernel-based algorithm.…”

The need for individualized studies to compare radiogenic second cancer (RSC) risk in proton versus photon Hodgkin Lymphoma patient treatments

“…Our adjustment of the photon dose to voxels below the 5% isodose line led to an average increase in our OAR mean dose by 14.5% in total breast, 2.6% in total lung, 4.6% for the thyroid. Previous papers reported an under dosage by the commercial TPS to voxels which fell below the 5% isodose line by 40 -60% 14,15 .…”

“…14,15 Instead, stray doses were determined using supplemental approaches. Specifically, stray radiation doses from photon therapy were calculated following the methodology from Howell et al 16 , which uses an analytical model to predict absorbed dose derived from TLD measurements of a HL IMRT treatment delivered to an anthropomorphic phantom.…”