Evaluating the influence of setup uncertainties on treatment planning for focal liver tumors

Abstract

Purpose: A mechanism has been developed to evaluate the influence of systematic and random setup variations on dose during treatment planning. The information available for studying these factors shifts from population-based models towards patient-specific data as treatment progresses and setup measurements for an individual patient become available. This study evaluates the influence of population as well as patient-specific setup distributions on treatment plans for focal liver tumors. Materials and Methods: 8 patients with focal liver tumors were treated on a protocol that involved online setup measurement and adjustment, as well as ventilatory immobilization. Summary statistics from these treatments yielded individual and population distributions of position at initial setup for each fraction as well as after setup adjustment. A convolution model for evaluation of the influence of random setup variation on calculated dose distributions has been previously described and investigated for application to focal liver radiotherapy by our department. Individual patient doses based on initial setup positions were calculated by applying the measured systematic offset to the initial treatment plan, and then convolving the calculated dose distribution with an anisotropic probability distribution function representing the individual patient’s random variations. A separate calculation with no offset and convolution using population averaged random variations was performed. Individual beam apertures were then adjusted to provide plans that ensured proper dose to the clinical target volume (CTV) following convolution with population distributions prior to and following setup adjustment. Results: Input distributions comprised 262 position measurements. Individual patient setup distributions for the course of treatment had systematic offsets ranging from (σ) 1.1 to 4.1 mm (LR), -2.0 to 1.4 mm (AP), and 5.6 to 1.7 mm (IS). Individual random setup variations (ƒã) ranged from 2.5 to 5.7 mm (LR), 2.1 to 8.3 mm (AP), and 4.1 to 10.8 mm (IS). The population random components were 4.0 mm (LR), 3.8 mm (AP), and 6.7 mm (IS) at initial setup. These were reduced to 2.1 mm, 2.3 mm, and 3.5 mm respectively following online setup correction. The initial, static, planned dose distribution overestimated the volume of liver irradiated to high doses, as inclusion of setup uncertainites generally blurred the resulting doses, shifting the higher dose region of normal liver DVHs to lower doses Further, the population-based dose convolution tended to predict a higher risk to the liver that the individual patient calculations. Plans generated to cover the CTV based on population convolutions using initial and corrected positions yielded very similar liver dose volume histograms (slight decrease of intermediate doses with the corrected setup distributions). Both plans showed significant reduction in liver high dose regions over the original static plan. For an individual plan, application of different individual random and systematic variations yielded Veff differences with a 3% range. Plan adjustment to account for random setup variations generally resulted in a lower Veff than initial planning using a PTV followed by calculation of delivered dose based on systematic and random offsets. Conclusion: This study hints at the factors that most strongly influence planning of liver treatments taking into account geometric variations. While not a complete picture, results indicate that systematic errors play a far more important role than random variations in dose to normal liver, in support of previous reports from other body sites. These data support the need for routine setup measurement (possibly offline) over the first few treatment fractions to remove systematic offsets. The importance of realistic incorporation of geometric variations as an initial step in treatment planning is demonstrated. This work was supported in part by National Cancer Institute grant 2 P01 CA59827.