Noncoplanar beams improve dosimetry quality for extracranial intensity modulated radiotherapy and should be used more extensively

Abstract

Arguing against the Proposition is David M. Shepard, Ph.D. Dr. Shepard obtained his M.S. and Ph.D. from the University of Wisconsin where he worked in the tomotherapy research group. He joined the faculty at the University of Maryland School of Medicine in 1999. In 2006 he moved to his current position as Director of Medical Physics, Swedish Health Services, Seattle, WA. He is certified by the American Board of Radiology in Therapeutic Radiologic Physics, is a fellow of the AAPM, and has served as President of the Northwest Chapter. Dr. Shepard's major research interests include developments in volumetric modulated arc therapy, image guided radiation therapy, motion management, and optimization, for which he has several patents and grants and has published over 30 papers in peer-reviewed journals. In principle, coplanar beam geometry is a subset of noncoplanar solution space and the latter should yield superior dosimetry. The dosimetric advantages of noncoplanar beam geometries have been clearly demonstrated for intracranial treatments via Gamma Knife and Linac machines.1 As a result, noncoplanar beams are systematically used for intracranial stereotactic radiosurgery treatment. The usefulness of noncoplanar beams in extracranial treatments, however, is less clear. I believe that differing utilization rates of noncoplanar beams in intracranial and extracranial treatments are due, not to the noncoplanar approach itself, but to the limited quality and quantity of noncoplanar beams applied to practical extracranial plans. Since hemispherical beam templates typically utilized in intracranial radiosurgery are not feasible for extracranial treatment, the beam orientation has to be selected. Unfortunately, manual selection of noncoplanar beam orientations is neither intuitive nor optimal. Automated beam orientation optimization has been previously researched, but more practical and robust beam orientation optimization algorithms to solve the complex integrated noncoplanar beam orientation and fluence optimization problem were not reported until recently. Breedveld et al.2 developed an automated beam orientation and optimization program, termed iCycle, to manage a large number of noncoplanar beams. They showed that noncoplanar plans consistently outperformed coplanar plans.3 More interestingly, using the same algorithm, Rossi et al.4 showed substantial dosimetric gains by increasing the number of noncoplanar beams from 12 to 25. Similarly, our group has optimized both the beam orientations and fluence maps for 12 lung SBRT cases using a 4π algorithm for both coplanar and noncoplanar plans.5 We have demonstrated that with fewer than ten beams, the difference in R50 (coplanar vs noncoplanar), defined as the ratio of the 50% isodose volume to that of the PTV, is insignificant and can be compensated for by using more coplanar beams, findings that have been previously observed. However, with more than 20 beams, R50 of the noncoplanar plan is 30% less than that of the coplanar plan and can no longer be matched using more coplanar beams. Using the same method, we showed that clinical plans using primarily coplanar beams could be meaningfully improved for lung and liver SBRT patients.5,6 For example, using 4π planning, the mean normal liver volume receiving <15 Gy was increased by 51 cm3 (range 21–107 cm3) with a 31% reduction of the mean normal liver dose, when compared against two partial arc VMAT plans. For lung SBRT patients, the critical organ doses were reduced by 32%–72%. The substantial improvement in critical organ sparing would allow a 40% target dose escalation. Even for the prostate that is centrally located, a significant reduction in the V50%, V80%, and V90% values for the rectum was achieved using 4π.7 Concerns about integral dose were alleviated by a recent study showing comparable noncoplanar and coplanar integral doses.8 The effect of longer x-ray paths in noncoplanar plans was offset by a shorter average beam entrance-to-ta