1

ProKnow CSI Plan Challenge

In my center, I have never personally done a plan for craniospinal irradiation (CSI).
Many years ago, I had only observed a CSI case during training, but assumed much has changed
since the 3D conformal days. So, I did some immediate research with my coworkers and my
radiation oncologist. I was then informed by my radiation oncologist that this treatment method
is specific to medulloblastoma cases, often in children prior to the well-developed intrathecal
chemotherapy techniques. My radiation oncologist led me to do some of my own research,
telling me to research it as if he had a case that needed treated for CSI. He did tell me to start that
he would prefer a supine method rather than the more common formerly used prone technique
due to patient comfort and re-producibility. So, in my research of current, combined with prior
methods, I chose to utilize the supine patient method with all 6MV photons planned with a
source-axis distance (SAD) treatment technique.
Using the patient oriented in the supine position, I first decided to create a block margin
in contouring mode on each, the brain planning target volume (PTV) and spine PTV based upon
the guidance by Parker et al1 of 1cm radially and craniocaudally. At this point, I set my patient
orientation be the 0, 0, 0 slice in line with the CT markers. I then calculated the collimator
rotation angle to be 11° for the brain field based upon the idea that I will be doing a maximum
field size on the spine field of 20 cm in the superior direction (arc tan(20/100)). With the
collimator rotation, I created the brain field in a half beam blocked fashion in the inferior
direction, covering the inferior extent of the C2 vertebral body and 1.5 cm superior, anterior and
posterior from the skin to apply appropriate flash coverage. This location of half beam blocked at
the C2 vertebral body placed my brain treatment isocenter at x = 0, y = -11.5, z = 0. The MLC
blocking was then applied along the PTV block margins of the brain and the spine, also
extending past the lenses and optic nerves (figure 1).
2

Figure 1. Right lateral beams eye view (BEV) of the brain displaying the chosen weight point
location as well as the field size and blocking of the critical structures.

This blocking method of the brain to the C2 vertebral body helped to protect the parotids and
submandibular glands as well. I decided to plan the brain as if I were doing a whole brain
technique for the clinic, so I chose to do forward planning field in field (FIF) technique to reduce
the hot spot while maintaining adequate PTV coverage. After completion, I normalized my plan
to receive 100% to the reference point of my weight point, which I placed ahead of time in the
most under dosed area of the brain clinical target volume (CTV). At this point, I was ready to
begin working on my spine fields.
I first decided to set my posterior spine fields based upon my centers former CSI
planning technique, which were 2 posterior fields at a gantry of 180° and a collimator and couch
angle both at 0°. With that, I created my superior spine field treatment isocenter at x = 0, y =
-31.5, z = 0, which is 20 cm away from the half beam blocked brain isocenter. I then extended
the field to 20 cm inferiorly and calculated my gap for the inferior spine field. With a depth of
6.5 cm and an inferior field of 20 cm with an additional second field extending 10 cm in the
3

superior direction, the gap was approximately 1 cm. With that information, I created my inferior
spine isocenter at x = 0, y = -62.5, z = 0 to account for the total of 30 cm from the fields and the
1 cm gap. My superior spine field was set to 5.3 x 40.0 and my inferior spine field was set to 5.9
x 12.7 in order to encompass the entirety of the Spine PTV. I then inserted MLCs on the fields
and created my blocking to fit along the Spine PTV block margin, allowing for adequate PTV
coverage and limiting excess dose to the lungs, heart, liver, kidneys and bowel. I kept the dose
normalized to the treatment isocenter of the superior spine field and created a plan sum of the
brain and spine plan. From the plan sum evaluation, in order to get 95% of the PTV covered by
36Gy, I normalized the spine plan to the 104% isodose line (IDL). I submitted my plan sum to
ProKnow and had a final score of 99.8/127, with my biggest issues being dose to the esophagus,
thyroid and target max to the spine PTV (Figure 2).

Figure 2. ProKnow score card of the plan sum between the Brain FIF plan and Spine 3D plan.
Seeing that I could obviously improve my scorecard in several areas, I was very pleased
with my brain FIF plan but decided to create another spine plan utilizing an inverse planning
intensity-modulated radiation therapy (IMRT) technique. Knowing that I would have to give the
4

optimizer appropriate target options, I decided to go into contour mode and create some
additional structures. But first, from my brain FIF plan, I created the 50% IDL into a structure in
order to establish a dose avoidance sector. In contour mode, I cropped my Spine PTV 1.0 cm
from the brain 50% IDL and Spine Block PTV at the brain 50% IDL. In planning mode, I created
my superior spine treatment isocenter at x = 0, y = -30.0, z = 0, which is 18.5 cm away from the
half beam blocked brain isocenter. I chose this isocenter placement due to the fact that the
optimizer would place a 1.0 cm extension on my largest lower objective (Figure 3).

Figure 3. Posterior BEV showing the 50% IDL brain structure (magenta), IMRT PTV spine
structure (cyan) and superior spine isocenter 18.5 cm from the brain isocenter, which is
represented by the 50% IDL brain structure.

I then created my inferior spine treatment isocenter at x = 0, y = -59.0, z = 0, which is 29.0 cm

away from the superior isocenter, allowing for appropriate PTV coverage at maximum field sizes
as needed. Using the guidance of field arrangement from Parker et al1, I decided to create 5
superior and 5 inferior spine fields, all 6MV and SAD at gantry positions 240°, 200°, 180°, 160°,
5

140° with all couch and collimator positions at 0°. The idea of these gantry positions was to
provide the optimizer an opportunity to gain adequate coverage to the spine PTV and limit dose
to the thyroid, esophagus, lungs, heart and bowel. After only 2 runs of the optimizer, which I
used to reduce my hot spot and esophageal dose, I was ready to create a plan sum with the brain
FIF plan. In the plan sum, I evaluated the dose needed to gain adequate PTV coverage, but keep
the hot spot under control. So, I kept my spine prescribed to 100% IDL with no plan
normalization. To my pleasant surprise, when submitting to ProKnow, I was able to achieve a
score of 123.85/127.0 (Figure 4).

Figure 4. ProKnow score card of the plan sum between the Brain FIF plan and Spine IMRT
plan.
After farther thinking and review, I would likely have created a contour between the gap
that I had created between the IMRT PTV spine contour and 50% IDL brain contour in order to
gain better coverage over the entirety of the spine PTV. However, in doing this, I would have
had to adjust my spine isocenters once again to allow for the 1 cm optimizer addition. With that,
I likely would have a hot spot develop in the spine and brain field control area, since that area
was where my overall hot spot had developed in the plan sum (Figure 5).
6

Figure 5. Transverse, coronal, sagittal and BEV display of the hot spot location in the IMRT
technique.

I am sure this hot spot developed due to the fact that I was trying to push dose to the area while
also limiting dose to the 50% IDL brain contour. So, I likely could have created rings and used
them as dose reduction as well as dose escalation.
Both plans have their issues, but I would definitely say that the IMRT method is much
better overall. If adjustments could be made with additional practice, I think the IMRT method
would be seamless. In both cases, I chose SAD technique to limit patient and couch movement
for setup accuracy, only making isocenter adjustments in the longitudinal direction. Though both
have adequate coverage for this plan challenge, the IMRT method was superior in hot spot
control as well as dose to the thyroid, esophagus, heart, lung D20% and submandibular glands.
However, to little surprise, the dose to the kidneys were slightly higher with the IMRT method,
but still within acceptable ranges (Figure 6a-b).
7

Figure 6a DVH display of the spine 3D (squares) and spine IMRT (triangles) methods with
brain FIF.

Figure 6b. Structure identification labels for the DVH in figure 6a.

With no surprise, the 3D spine method had a hot spot inferiorly where the fields
overlapped with exit dose past the gap point anteriorly, also creating underdosing in that area of
the spine PTV; whereas the IMRT spine method did a nice job of hot spot control in the field
junction area, while also maintaining coverage to the PTV (Figure 7).
8

Figure 7. Transverse, coronal and sagittal view of the gap point between the superior and
inferior spine fields for the 3D spine (left) and IMRT spine (right) plans.

Conversely, the 3D spine plan did a better job than the IMRT spine plan at maintaining dose to
the spine PTV at the junction of the superior brain fields (Figure 8). However, this is likely due
to a slightly inadequate contouring step on my part that will be adjusted with future CSI planning
techniques.
9

Figure 8. Transverse, coronal and sagittal view of the match point between the superior brain
and inferior spine fields for the 3D spine (left) and IMRT spine (right) plans.

This plan challenge was very interesting and I really enjoyed learning the older methods
while applying the new techniques. This challenge has actually challenged me a lot more to do
my own research on techniques and apply the best method based upon various studies. I will
certainly utilize the beam arrangement used in the IMRT method more often, especially on
various spine cases that may benefit from dose reduction to the thyroid and esophagus. I also
enjoyed utilizing different mathematical techniques that we often take for granted due to the
intelligence of our planning systems, like a gap calculation and collimator rotation calculation.
This was a great plan challenge and I was very happy with the result, making me more confident
in my planning abilities, especially if we are to ever have a CSI case once again in the clinic.
10

References
1. Parker W, Filion E, Roberge D, Freeman CR. Intensity-modulated radiotherapy for
craniospinal irradiation: target volume considerations, dose constraints, and competing
risks. Int J Radiat Oncol Biol Phys. 2007;69(1):251-257.
doi:10.1016/j.ijrobp.2007.04.052