Effects of tissue heterogeneity and comparisons of collapsed cone and Monte Carlo fast neutron patient dosimetry using the University of Washington clinical neutron therapy system (CNTS).

Fast neutron therapy is a high linear energy transfer (LET) radiation treatment modality offering advantages over low LET radiations. Multileaf collimator technology reduces normal-tissue dose (toxicity) and makes neutron therapy more comparable to MV x-ray treatments. Published clinical-trial and other experiences with fast neutron therapy are reported. Early comparative studies failed to consider differences in target-dose spatial conformality between x-ray and neutron treatments, which is especially important for organs-at-risk close to tumor targets. Treatments planning systems (TPS) for high-energy neutrons lag behind TPS tools for MV x-rays, creating challenges for comparative studies of clinical outcomes. A previously published Monte Carlo model of the University of Washington (UW) Clinical Neutron Therapy System (CNTS) is refined and integrated with the RayStation TPS as an external dose planning/verification tool. The collapsed cone (CC) dose calculations in the TPS are based on measured dose profiles and output factors in water, with the absolute dose determined using a tissue-equivalent ionization chamber. For comparison, independent (external) Monte Carlo simulation computes dose on a voxel-by-voxel basis using an atlas that maps Hounsfield Unit (HU) numbers to elemental composition and density. Although the CC algorithm in the TPS accurately computes neutron dose to water compared to Monte Carlo calculations, calculated dose to water differs from bone or tissue depending largely on hydrogen content. Therefore, the elemental composition of tissue and bone, rather than the material or electron density, affects fast neutron dose. While the CC algorithm suffices for reproducible patient dosimetry in fast neutron therapy, adopting methods that consider tissue heterogeneity would enhance patient-specific neutron dose accuracy relative to national standards for other types of ionizing radiation. Corrections for tissue composition have a significant impact on absolute dose and the relative biological effectiveness (RBE) of neutron treatments compared to other radiation types (MV x-rays, protons, and carbon ions).

Scattering kernels for fast neutron therapy treatment planning.

The University of Washington (UW) Clinical Neutron Therapy System (CNTS) has been used to treat over 3300 patients. Treatment planning for these patients is currently performed using an MV x-ray model in Pinnacle® adapted to fit measurements of fast neutron output factors, wedge factors, depth-dose and lateral profiles. While this model has provided an adequate representation of the CNTS for 3D conformal treatment planning, later versions of Pinnacle did not allow for isocentric gantry rotation machines with a source-to-axis distance of 150 cm. This restriction limited the neutron model to version 9.0 while the photon and electron treatment planning at the UW had moved on to newer versions. Also, intensity modulated neutron therapy (IMNT) is underdevelopment at the UW, and the Pinnacle neutron model developed cannot be used for inverse treatment planning. We have used the MCNP6 Monte Carlo code system to develop Collapsed Cone (CC) and Singular Value Decomposition (SVD) neutron scattering kernels suitable for integration into the RayStation treatment planning system. Kernels were generated for monoenergetic neutrons with energies ranging from 1 keV to 51 MeV, i.e. the energy range most relevant to the CNTS. Percent depth dose (PDD) profiles computed in RayStation for the CC and SVD kernels are in excellent agreement with each other for depths beyond the beam's dose build-up region (depths greater than about 1.7 cm) for open 2.8 × 2.8 cm2, 10.3 × 10.3 cm2, and 28.8 × 32.8 cm2 fields. Lateral profiles at several depths, as well as the PDD, calculated using the CC kernels in RayStation for the full CNTS energy spectrum pass a 3%/3 mm gamma test for field sizes of 2.8 × 2.8 cm2, 10.0 × 10.3 cm2, and 28.8 × 32.8 cm2.

Dosimetric characteristics of the University of Washington Clinical Neutron Therapy System.

The University of Washington (UW) Clinical Neutron Therapy System (CNTS), which generates high linear energy transfer fast neutrons through interactions of 50.5 MeV protons incident on a Be target, has depth-dose characteristics similar to 6 MV x-rays. In contrast to the fixed beam angles and primitive blocking used in early clinical trials of neutron therapy, the CNTS has a gantry with a full 360° of rotation, internal wedges, and a multi-leaf collimator (MLC). Since October of 1984, over 3178 patients have received conformal neutron therapy treatments using the UW CNTS. In this work, the physical and dosimetric characteristics of the CNTS are documented through comparisons of measurements and Monte Carlo simulations. A high resolution computed tomography scan of the model 17 ionization chamber (IC-17) has also been used to improve the accuracy of simulations of the absolute calibration geometry. The response of the IC-17 approximates well the kinetic energy released per unit mass (KERMA) in water for neutrons and photons for energies from a few tens of keV up to about 20 MeV. Above 20 MeV, the simulated model 17 ion chamber response is 20%-30% higher than the neutron KERMA in water. For CNTS neutrons, simulated on- and off-axis output factors in water match measured values within ~2% ± 2% for rectangular and irregularly shaped field with equivalent square areas ranging in a side dimension from 2.8 cm to 30.7 cm. Wedge factors vary by less than 1.9% of the measured dose in water for clinically relevant field sizes. Simulated tissue maximum ratios in water match measured values within 3.3% at depths up to 20 cm. Although the absorbed dose for water and adipose tissue are within 2% at a depth of 1.7 cm, the absorbed dose in muscle and bone can be as much as 12 to 40% lower than the absorbed dose in water. The reported studies are significant from a historical perspective and as additional validation of a new tool for patient quality assurance and as an aid in ongoing efforts to clinically implement advanced treatment techniques, such as intensity modulated neutron therapy, at the UW.

MCNP6 model of the University of Washington clinical neutron therapy system (CNTS).

A MCNP6 dosimetry model is presented for the Clinical Neutron Therapy System (CNTS) at the University of Washington. In the CNTS, fast neutrons are generated by a 50.5 MeV proton beam incident on a 10.5 mm thick Be target. The production, scattering and absorption of neutrons, photons, and other particles are explicitly tracked throughout the key components of the CNTS, including the target, primary collimator, flattening filter, monitor unit ionization chamber, and multi-leaf collimator. Simulations of the open field tissue maximum ratio (TMR), percentage depth dose profiles, and lateral dose profiles in a 40 cm × 40 cm × 40 cm water phantom are in good agreement with ionization chamber measurements. For a nominal 10 × 10 field, the measured and calculated TMR values for depths of 1.5 cm, 5 cm, 10 cm, and 20 cm (compared to the dose at 1.7 cm) are within 0.22%, 2.23%, 4.30%, and 6.27%, respectively. For the three field sizes studied, 2.8 cm × 2.8 cm, 10.4 cm × 10.3 cm, and 28.8 cm × 28.8 cm, a gamma test comparing the measured and simulated percent depth dose curves have pass rates of 96.4%, 100.0%, and 78.6% (depth from 1.5 to 15 cm), respectively, using a 3% or 3 mm agreement criterion. At a representative depth of 10 cm, simulated lateral dose profiles have in-field (⩾ 10% of central axis dose) pass rates of 89.7% (2.8 cm × 2.8 cm), 89.6% (10.4 cm × 10.3 cm), and 100.0% (28.8 cm × 28.8 cm) using a 3% and 3 mm criterion. The MCNP6 model of the CNTS meets the minimum requirements for use as a quality assurance tool for treatment planning and provides useful insights and information to aid in the advancement of fast neutron therapy.