Kyle Garafolo

Dos 773 – Clinical Practicum III

October 16, 2019

Craniospinal Irradiation Plan Study

Introduction
Craniospinal irradiation (CSI) is often considered to be a highly complex treatment
planning process. Part of the complexity of this planning method is due the numerous organs-at-
risk (OAR) near the volume to be treated, as well as the large (and typically matching) treatment
fields. Craniospinal irradiation is usually recommended for the treatment of medulloblastoma
and other rare tumors with leptomeningeal disease, such as germ-cell tumors, atypical teratoid
rhabdoid tumors, and ependymomas.1 The CSI case I planned involved a male patient treated in
the supine position. The radiation prescription was written for a total dose of 36 Gy to be treated
for 20 fractions at 1.8 Gy per fraction. Treatment planning was performed using Varian Eclipse
treatment planning system (TPS) version 15.6 and delivered on a Varian TrueBeam linear
accelerator.

Patient positioning, setup, and treatment fields

Patients requiring CSI treatments can be positioned either supine or prone. Research by
Tai et al2 found that the supine position offers several advantages over the prone position. Some
of these advantages include: superior patient comfort, improved anesthesia access (when
necessary) and anesthetic complication rates, easier patient positioning and stability, and more
simplistic treatment planning options. For these reasons, I chose to treat the patient in this study
in the supine, head-first position with his arms positioned at his sides. A thermoplastic mask was
used to aid in patient immobilization and setup reproducibility.
Due to the risk for cerebrospinal fluid (CSF) seeding, CSI treatments will include the
entire CSF-bearing anatomy.3 Treatment volumes should include the entire brain and intracranial
CSF spaces, as well as the entire spinal canal inferiorly to include the end of the thecal sac. The
thecal sac will be located at the approximate level of second or third sacral vertebra, as
confirmed on MRI. Based on these treatment volumes, the nearby OAR that will be evaluated
include the heart, kidneys, liver, lungs, lenses, optic nerves, parotid glands, submandibular
glands, thyroid gland, esophagus, and the entire bowel space.
 There has been much research conducted on the various techniques used for CSI
treatments. Conventional techniques employ the use of 3-dimensional conformal radiation
therapy (3D-CRT), while modern options include intensity-modulated radiation therapy (IMRT),
volumetric-modulated arc therapy (VMAT), TomoTherapy, and proton pencil beam scanning
(PBS).1 Modern treatment techniques have proven advantageous over 3D-CRT in achieving
superior dosimetric results, particularly in terms of conformity/homogeneity indices as well as
dose to nearby OAR (Figure 1).1 Based on the literature I reviewed and the options I had
available, I opted to create a CSI treatment plan utilizing the VMAT technique.

Figure 1. From left to right, comparison of dose distribution for 3D-CRT, IMRT, VMAT, TomoTherapy,
and proton PBS planning techniques.1

Using the arc geometry tool within Eclipse TPS, I opted to use 3 isocenters spaced
equidistant along the brain/spinal axis (Figure 2). Beginning at the brain isocenter, each
subsequent isocenter was spaced exactly 23 cm inferiorly to allow for easier shifting for the
radiation therapists. At each isocenter, I used 2 full arcs (179°-181° and 181°-179°) and slight
collimator angles (10° and 350°) to allow the optimizer more freedom to modulate the beams
while minimizing inter-leaf leakage (Figure 3). Field sizes were automatically established by the
arc geometry tool based on my setting of 0.5 cm margin to the PTV edge. Each of the 6 fields
were planned using an energy of 6 MV and a source-to-axis distance (SAD) of 100 cm in order
to minimize any potential for gantry collision during arc rotations. No plan normalization was
used as the plan was created volumetrically.

Figure 2. Lateral and coronal views of 3 isocenters (brain, upper spine, and lower spine) and field
arrangements.

Figure 3. Field parameters highlighting gantry rotation, collimator angles, and equidistant shifts between
isocenters
Treatment planning process and optimization
Prior to entering the photon optimizer, it was necessary to create several structures
specific for optimization. I started by creating a combined PTV of the brain and spine structures,
labeled PTV_Total. I then created PTV_Brain_Opti and PTV_Spine_Opti structures by adding a
0.1 mm margin to each PTV, adding a smoothing effect, then cropping one from the other to
avoid overlap. Any region where these optimization structures were way over/under covering the
PTV_Total was corrected. In areas near the lenses, optic nerves, and kidneys, the 1 mm
expansion was not applied in order to make it easier to achieve dose constraints to these OAR.
Because there was a small area of overlap of the PTV_Brain with the optic nerves, I created an
overlap structure for the optic nerves that only included these overlapping regions plus a 1 mm
margin, labelled PTV_Optic_Nerve_OL. The PTV_Optic_Nerve_OL structure was then cropped
from the PTV_Brain_Opti to make separate PTVs within the brain that I could then individually
control dose to. I was then able to create an avoidance structure for the portion of optic nerves
that did not overlap with the PTV_Brain_Opti. By separating the optic nerves into overlap and
avoid structures, I was able to better control the dose to the optic nerves both within and outside
the PTV_Brain_Opti structure.
After verifying all my fields, prescription, and other setup parameters were appropriate, I
was then able to enter the photon optimizer. I began by setting the Normal Tissue Objective
(NTO) to automatic with a priority of 150 and ensured jaw tracking was enabled for treatment. I
then began removing structures from the optimization list that were either far from the PTV
structures or could easily achieve the dose constraints per the ProKnow software. This made the
list of structures to optimize on easier to visualize. Next, I added upper and lower objectives to
my PTV structures, including the PTV_Brain_Opti, PTV_Spine_Opti, PTV_Spine, and
PTV_Optic_Nerve_OL. I then continued to add either upper, mean, or upper gEUD objectives
on the nearby OAR depending on plan metrics found in ProKnow.
For my PTV_Opti structures, I start by entering the upper 0% objective to receive 103-
105% of the prescription dose with a priority of 135; the lower 100% objective value is set to
receive 101% of the prescription dose with a priority of 135. For PTV overlap structures, I
typically enter the upper 0% objective to receive 101-102% of the prescription dose and the
lower 100% objective to receive 100% of the prescription dose, both with priorities of 135. This
way, I can help minimize hot spots from appearing in the PTV overlap structures. However,
given that the PTV overlap structure in this plan were the optic nerves which had lower dose
objectives than the PTV itself, I had to reduce these objectives to lower the maximum dose
received by the optic nerves, all while still maintaining adequate coverage to the PTV_Brain. For
the OAR, I typically placed either upper, mean, or upper gEUD (a = 1) objectives with a priority
of 80 on these structures, mainly based off the dose criteria provided within the ProKnow
software. Depending on how the plan is optimizing, these values were modified in order to either
provide better target coverage or reduce dose to the nearby OAR. My final optimization values
can be found in figure 4 below.

Figure 4. Final optimization objectives

Plan evaluation
After numerous planning attempts, I was finally satisfied with my plan outcome.
However, my final plan did not come without modifications along the way. Initial planning
attempts proved challenging at meeting the dose constraints for the lenses, optic nerves, and
kidneys. Due to the fact that part of the optic nerves were included in the PTV_Brain and had
lower dose objectives, I found it challenging to meet both the optic nerve objectives and achieve
proper coverage to the PTV. In order to help lower the dose to the lenses and optic nerves, I
placed a 60° avoidance sector from 330°-30° on one of my brain fields (Figure 5). In conjunction
with pushing the optimization objectives more, this technique was able to reduce dose to these
OAR to an acceptable level while allowing the optimizer to maintain coverage to the
PTV_Brain.

Figure 5. Avoidance sector used to minimize dose received to the lenses and optic nerves.

After each planning attempt, I uploaded my dose and plan files into ProKnow where I
could see my plan metrics and scoring. Based on these scores, I would re-optimize and make
minor objective/priority adjustments until I achieved an outcome that I deemed satisfactory. My
final plan score was 126.49 / 127, with only a few structures barely not meeting the ideal
outcome metrics (Figure 6). I was quite pleased with the final plan score as I knew I had
exhausted several optimization methods along the way.

Figure 6. Final ProKnow plan metric scoresheet

My final plan had a maximum dose of 110.4% (39.73 Gy), which was located in the bone
on the lateral aspect of the cranium (Figure 7). I found this location to be acceptable given that it
was located within the PTV away from any OAR. Throughout the entire PTV, my treatment plan
did not really have many hot or cold spots. Most noticeably, however, were a few cold spots at
the approximate level of the optic nerves (Figure 8). I found these cold spots to be reasonable
given how hard I had to push on the optimizer to “carve out” dose near the optic nerves. Other
than these few areas, I found my plan to be rather conformal and homogenous throughout
(Figures 9 & 10). Dose volume histograms (DVH) for both the cranial/upper spinal and lower
spinal structures can be found below (Figures 11 & 12).

Figure 7. Area of maximum dose found within the PTV.

Figure 8. Isodose line and color wash depiction of dose being “carved out” near the optic nerves.
Figure 9. Isodose line and color wash depiction of lateral aspect of entire brain and spine PTV.
Figure 10. Isodose line and color wash depiction dose avoiding the kidneys.
Figure 11. Cranial / upper spinal DVH
Figure 12. Lower spinal DVH
Reflection
This CSI plan study really made me think outside the box when it came to designing the
treatment plan. Not only did I learn a new planning technique, I also learned how to add
avoidance sectors as well as multiple isocenters for VMAT planning/optimization. Prior to
conducting research, I was intending on creating a traditional 3D-CRT plan using lateral fields
for the brain and PA fields for the spine. However, after reading current literature, I quickly
realized the benefits that VMAT can offer over 3D-CRT for CSI cases. This CSI plan study has
shown the ability of VMAT to achieve a highly conformal and homogenous treatment plan while
minimizing dose to nearby OAR as much as possible. Of note, however, this planning technique
does increase the integral dose to surrounding tissues and organs that may increase the risk for
future secondary malignancies.4 For these reasons, extreme care must be taken when determining
the best planning technique to use on an individual patient basis.
 References

1. Seravalli E, Bosman M, Lassen-Ramshad Y, et al. Dosimetric comparison of five

different techniques for craniospinal irradiation across 15 European centers: Analysis on
behalf of the SIOP-E-BTG (radiotherapy working group). Acta Oncologica.
2018;57(9):1240-1249. https://doi.org/10.1080/0284186X.2018.1465588
2. Tai P, Koul R, Vu K, et al. A simplified supine technique expedites the delivery of
effective craniospinal irradiation to medulloblastoma – comparison to other techniques in
the literature. Cureus. 2015;7(12):e404. https://doi.org/10.7759/cureus.404
3. Washington CM, Leaver D. Principles and practice of radiation therapy. 4th ed. St
Louis, MO: Elsevier Mosby; 2016.
4. Fathy S, Alalawi T, Al-Dhaibani N. Treatment planning evaluation of volumetric
modulated arc therapy (VMAT) for craniospinal irradiation (CSI). Int J Radiat Oncol
Biol Phys. 2017;99(2):e658. https://doi.org/10.1016/j.ijrobp.2017.06.2189