IMRT Dosimetry of the Head and Neck:

1
IMRT Dosimetry of the Head and Neck A
Comparison of Treatment Plans Using Linac-Based
IMRT and Helical Tomotherapy Ke Sheng, Ph.D.,
Janelle A. Molloy, Ph.D., and Paul W. Read,
Ph.D., M.D. University of Virginia,
Charlottesville, VA Mayo Clinic,
Rochester, MN
2
Purpose/Objective
To date, most IMRT delivery has occurred
using linear accelerators (linacs), although
helical tomotherapy has become commercially
available. In practice, linac-based IMRT has
potential advantages in its ability to deliver
non-coplanar beams and dimensionally precise
beamlets. Helical tomotherapy has the ability to
deliver many more beamlets but can not deliver
non-coplanar beams. Herein, we report the
findings of a comparison of linac-based and
helical tomotherapy-based treatment plans for
IMRT of certain sites in the head and neck.
3
Methods
20 head and neck patients were selected from
our database. These had received treatment using
our departments linac-based IMRT system. This
consisted of inverse treatment planning using
Corvus and step and shoot delivery on a Varian
2300 C/D multileaf collimator with a 1 cm beamlet
size. 5 patients each were chosen who had tumors
in the base-of-tongue (BOT), tonsil (T), sinus
and nodes (SN), and sinus only (S). All
patients were simulated on a Philips Picker,
PQ5000 CT scanner. The clinical target volume
(CTV) included the gross target volume (GTV) with
an expansion of 1-2 cm. The PTV was produced by
adding a 3 mm margin around the CTV in all
dimensions. All target volumes and normal
structures were contoured on AQsim by a single
radiation oncologist specializing in treatment of
head and neck disease. Normal structures included
the brainstem, oral cavity, mandible, parotids,
larynx, spinal cord and spinal cord buffer. 50
Gy in 25 fractions was prescribed. The CT images
and structure sets were transferred to the Corvus
inverse treatment planning system. These
identical structure/image sets were then
transferred from Corvus to the Tomotherapy
planning system to ensure consistency.
4
Methods, continued
Corvus Planning Linac based treatment plans
were generated using the following beam
arrangements for each tumor site. Base of
tongue 6 coplanar, 1 sagittal (ant/inf)
Tonsil 7 coplanar fields Sinus and Nodes 7
coplanar Sinus only 6 coplanar, 2-3
sagittal Normal structure constraints were
optimized to minimize the dose to these
structures without compromising the PTV coverage
with at least 95 of the PTV receiving the
minimum prescribed dose.
5
Methods, continued
Tomotherapy Planning In the tomotherapy
optimization, a 2.5 cm field width was used.
Other common parameters were pitch 0.3 and
nominal modulation factor (i.e., ratio of maximum
to average beamlet intensity) of 2.5. Penalties
were given to specify the PTV minimum dose and
the OAR maximum dose. The optimization is an
iterative process with a series of interactions
between the operator and the software to achieve
a stable and satisfactory result. The number of
iterations during this optimization process
varies but is in the range of 60 to 150 per plan.
6
Methods, continued
Dose point analysis Plans were compared by
noting the superiority of relevant dosimetric
parameters for the two planning systems. We
defined significant superiority as a difference
that exceeded 10. Dosimetric parameters include
the maximum and minimum planning target volume
(PTV) dose, and the maximum and mean dose to
organs at risk (OARs). Data was analyzed by
comparing these parameters on a case-by-case
basis for the linac and tomotherapy-based plans.
An analysis of the effect of the target volume
was performed by defining a T/L ratio as the
number of dosimetric parameters that were
superior for the tomotherapy-based dosimetry to
those that were superior for the linac-based
dosimetry.
7
Methods, continued
DVH and EUD/NTCP analysis We performed an
extended analysis on the 10 base-of-tongue and
tonsil patients. DVH data from Corvus and
Tomotherapy were extracted for comparison. EUD
and NTCP calculation was based on the SDR model
(1), which is defined by
and EUD is defined by
where is the ith unit volume, Di is the dose to
that unit volume, n is the volume factor, m is
the gradient and is the dose to cause
complication to single functional unit. 1.
Burman, C., G. J. Kutcher, et al. (1991).
"Fitting of normal tissue tolerance data to an
analytic function." Int J Radiat Oncol Biol Phys
21(1) 123-35.  
8
Results
Both Tomotherapy and Corvus plans show good
coverage of the PTV with greater than 95 percent
of the PTV receiving 50 Gy. The difference of
these 2 delivery modalities is shown in Figure 1
and 2, where tomotherapy plans have a much
steeper PTV DVH that indicates greater dose
homogeneity inside the PTV. The higher
homogeneity of the tomotherapy plans is also
represented by a lower standard deviation. The
average standard deviation of 10 tomotherapy
plans is 0.41Gy and the average standard
deviation of the Corvus plans is 1.45 Gy.
9
Results
The normal tissue dose comparison by DVH for
BOT and Tonsil patients is exhibited in figure 3
and 4, where 5 organs are listed with left and
right parotids in separate figures. The combined
DVH of five patients shows reasonable similarity
in terms of maximum dose and shape. The
difference between Corvus and Tomotherapy plans
is better visualized on the combined DVHs in the
third column, where a lower dose can be observed
for all OARs with Tomotherapy. The quantitative
dose difference is also represented in the EUD
and NTCP calculation. The average EUD reduction
is 17.6 for BOT patients and 27.14 for tonsil
patients. NTCP of Corvus and tomotherapy plans
are calculated respectively for all OARs.
Tomotherapy plans showed a significant reduction
of the probability of complication of parotids
for both BOT and tonsil patients (78.5 vs 86.5
respectively). The mean dose to the parotids for
all 10 patients is 18.3 Gy and 12.6 Gy for Corvus
and Tomotherapy plans respectively.
10
Results, continued
Fig 1. PTV DVHs for 5 BOT cancer patients are
plotted individually for Corvus plans in a,
Tomotherapy plans in b, and average PTVs for each
system in c.
Fig 2. PTV DVHs for 5 tonsil cancer patients are
plotted individually for Corvus plans in a,
Tomotherapy plans in b, and average PTVs for each
system in c.
11
Results, continued
Fig 3. DVHs for 5 BOT cancer patients are plotted
as individual OAR DVHs for Corvus plans in a,
Tomotherapy plans in b, and as average Corvus and
Tomotherapy DVHs in c..
12
Results, continued
Fig 4. DVHs for 5 BOT cancer patients are plotted
as individual OAR DVHs for Corvus plans in a,
Tomotherapy plans in b, and as average Corvus and
Tomotherapy DVHs in c.
13
Results, continued
Table 2. EUD for BOT
Table 3. EUD for Tonsil
14
Results, continued
Table 4. NTCP for BOT plans
Table 5. NTCP for Tonsil plans
15
Results, continued
Fig. 5 A composite Tomotherapy plan based on 10
patients.
16
Conclusions
Helical tomotherapy-based treatment planning
for head and neck IMRT yields significant
dosimetric improvements over certain linac-based
plans for targets in the base-of-tongue, tonsil,
and sinus plus nodes. Linac-based treatment
plans were equivalent to tomotherapy plans for
sinus only targets. The ability to deliver
non-coplanar beams on a linac does not compensate
for the improved dosimetry available from the
tomotherapy planning system.
17
Conclusions
The OARs have better avoidance with a
general lower DVH and an average EUD reduction of
17.4 for BOT patients and 27.14 for tonsil
patients. NTCP shows a larger improvement
although the probability numbers are unimportant
for a 50 Gy prescription dose. As far as the most
concerned problem in oropharyngeal carcinoma
treatment, the mean risk of xerostomia is reduced
for 30 if it is linear with mean dose, or is
reduced for 82 if using SDR model and 25 Gy as
TD50.