ESTR

O 2

01

7

Poster presented at:

Stereotactic body radiation therapy treatment planning using target volume partitioning

James L Robar, PhD, FCCPMDepartment of Radiation Oncology, Dalhousie University

Department of Physics and Atmospheric Science, Dalhousie UniversityHalifax, Nova Scotia, Canada

ObjectivesIn this work we describe and evaluate a novel approach to StereotacticBody Radiation Therapy (SBRT) treatment planning for the spine usingVolumetric Modulated Arc Therapy (VMAT) involving partitioning of thePlanning Target Volume (PTV) prior to optimization. In this approach, sub-PTV volumes are defined that minimize concavity, thereby allowingprioritization of spinal cord sparing by the optimizer. We compare this newapproach to the status-quo method in our centre with using standard planquality and efficiency metrics.

Methods

Results

Conclusions

VMAT is an effective planningapproach for SBRT of the spine[1-3] but the vertebra presents achallenging target volume to theoptimizer. While competingpriorities are set for PTVcoverage and spinal cord sparing,the optimizer is also constrainedto spare normal tissues inconcavities surrounding spinousand transverse processes. Inthis study, we evaluated a noveltreatment planning approach(Spine SRS Element, BrainlabAG) that partitions the PTV intosimpler sub-volumes (Figures 1and 2) whereby the total amountof concavity is minimized. Eachsub-volume is then assigned to aseparate arc for VMAToptimization and delivery.

We evaluated this approach bycomparing it to the standardplanning method in our centre(RapidArc, Eclipse 11, VarianMedical Systems) for eightsample cases (Figure 3). In bothsystems, co-planar arcs wereused. In Eclipse, two coplanararcs were defined, while inElements, this baseline numberof arcs is multiplied by thenumber of PTV sub-volumesgenerated (typically four). Inboth systems, 24 Gy/2# wasprescribed to the PTVencompassing 90% isodosesurface. The PRV cord structure(approximately equivalent to thethecal sac) was limited to amaximum dose of 17 Gy, with 10Gy to 10% of its volume. Inboth systems, default normaltissue optimization settings wereused. Treatment plans wereevaluated with regard to PTVcoverage, PTV dose homogeneity(10% volume hotspot), inversePaddick conformity index,gradient distance, i.e., distancealong AP axis for dose to fallfrom 24 to 12 Gy toward thespinal cord, maximum dose tospinal cord, volume of spinal cordreceiving 10 Gy, and total MUs.

All treatment plans provided coverage to >98% of the PTV volume by the prescriptiondose level. The PTV 10%V hotspot was not significantly different between Eclipse andSpine SRS Element plans. The inverse Paddick index was statistically superior for theSpine SRS Element plans (Wilcoxan p=0.002, Figures 5, 10). On average, Spine SRSElement gave an improved dose gradient toward the spinal cord (Figures 6 and 10) butthis did not reach significance (p>0.3). Significant reduction of the volume of the spinalcord receiving 10 Gy was observed with the Spine SRS Element (p=0.05, Figure 7). Themaximum dose received by the spinal cord was equivalent for both techniques (Figure 8).No significant difference was observed with regard to required Monitor Units (Figure 9).

References

The PTV partitioning approach segments the complex vertebral PTV intosimplistic sub-volumes, allowing the spinal cord sparing to be prioritized duringVMAT optimization. In the cases studied here, two plan quality metrics wereimproved significantly: conformity of the prescription isodose surface andsparing of the spinal cord at the 10 Gy level. While PTV partitioning involves amultiplication of the number of VMAT arcs required, no increase in totalMonitor Units was observed.

Figure 1. A k-means algorithm partitionsthe spine PTV into 8 sectors. Adjacentsectors are then combined to form sub-volumes of the PTV that minimize concavity(shown in red). In this example, theselection in the lower figure is preferable tothat in the upper figure. The processcontinues as in Figure 2. The goal of thispre-optimization step is to generate typicallyfour sub-volumes, each of which placesminimal constraint on the VMAT optimizer.

ESTRO � Vienna � 2017

Figure 2. The recombination procedure continues, minimizingconcavity in each sub-volume. Each sub-volume will then be assignedto a separate arc for treatment delivery.

☑

Figure 3. Eight sample spinal target volumes were used forcomparative treatment planning.

Figure 4. For all plans, co-planar arcs of 358 degrees were used. Forthe Spine SRS Element plans, the total number of arcs is multiplied bythe number of sub-volumes. A prescription of 24 Gy in 2 fractions wasapplied to the covering isodose surface. A single spine PRV was usedduring VMAT optimization in Elements, with Dmax of 17 Gy, with nomore than 10 Gy allowable to 10% of the PRV volume.

Spinal cord %V receiving 10 Gy Spinal cord Dmax Monitor Units

Inverse Paddick index Gradient distance

Figure 5. Inverse Paddick index was improved bySpine SRS Element (p=0.002).

Figure 6. Distance over which dose falls from24 to 14 Gy in AP plane toward the spinal cord.Improvement with Spine SRS Element wasobserved but was non-significant (p>0.3).

Figure 7. The volume of spinal cord receiving10 Gy was reduced with Spine SRS Element(p=0.05).

Figure 8. The reduction of spinal cordmaximum dose observed with Spine SRSElement was non-significant (p>0.1).

Figure 9. Monitor units were notsignificantly different between the twoplanning techniques.

1. Pokhrel D, et al, On the use of volumetric-modulated arc therapy for single-fraction thoracic vertebral metastases stereotactic body radiosurgery, Med Dosim 2017, 42(1):69-75.

2. Zach L, et al, Volumetric Modulated Arc Therapy for Spine Radiosurgery: Superior Treatment Planning and Delivery Compared to Static Beam Intensity Modulated Radiotherapy, Biomed Res Int, 2016.

3. Woo, QJ et al, Volumetric arc intensity-modulated therapy for spine body radiotherapy: comparison with static intensity-modulated treatment, Int J Radiat BiolPhys 2009 75(5):1596-604.

Figure 10. Example of reduction gradient distance and improvement ofconformity of the prescription isodose surface.

EP-1520James Robar DOI: 10.3252/pso.eu.ESTRO36.2017

Physics track: Treatment plan optimisation: algorithms