Three-dimensional versus four-dimensional dose calculation for volumetric modulated arc therapy of hypofractionated...

Respiratory motion is a non-negligible source of uncertainty in radiotherapy. A common approach is to delineate the target volume in all respiratory phases (ITV) and to calculate a treatment plan using the average reconstruction of the four-dimensional computed tomography (4DCT) scans. In this study the extent of the interplay effect caused by interaction between dynamic dose delivery and respiratory tumor motion, as well as other motion effects were investigated. These effects are often ignored when the ITV concept is used.

Nine previously treated patients with in ten abdominal or thoracic cancer lesions (3 liver, 3 adrenal glands and 4 lung lesions) were selected for this planning study. For all patients, phase-sorted respiration-correlated 4DCT scans were taken, and volumetric modulated arc therapy (VMAT) treatments were planned using the ITV concept. Margins from ITV to planning target volume (PTV) of 3-10mm were used. Plans were optimized and dose distributions were calculated on the average reconstruction of the 4DCT. 4D dose distributions were calculated to evaluate motion effects, caused by the interference of dynamic treatment delivery with respiratory tumor motion and inhomogeneously planned target dose. These calculations were performed on the phase-sorted CT series with a respiration-correlated assignment of the treatment plan's monitor units (MU) to the respiration phases of the 4DCT. The 4D dose was accumulated with rigid as well as deformable registrations of the CT series and compared to the original 3D dose distribution. Maximum, minimum and mean doses to ITV and PTV, and maximum or mean doses to organs at risk (OAR), were compared after rigid accumulation. The dose variation in the gross tumor volume (GTV) was compared after deformable registration.

Using rigid registrations, variations in the investigated dose parameters between 3D and 4D dose calculations were found to be within -2.1% to 1.4% for all target volumes and within -0.8% to 1.7% in OAR. Using deformable registrations, dose differences in the GTV were below 3.8% for dose accumulation of lung and adrenal gland patients. For liver patients the used deformable registrations were not considered to be robust enough. It was also shown that a major part of the dose differences originates from the Hounsfield unit differences between 3D and 4D calculations, regardless of the interplay effect.

The evaluated motion effects during VMAT treatments resulted in negligible dose variability. Therefore, the approximation of calculating the dose on the average reconstruction of the 4DCT (3D dose calculation), instead of calculating on the respiration-correlated phase CTs (4D dose calculation) with assignment of the corresponding MUs, gives acceptable results.