Professional Documents
Culture Documents
2012 Dosimetric Aspects of Inverse-Planned Modulated-Arc Total-Body Irradiation - Held2012
2012 Dosimetric Aspects of Inverse-Planned Modulated-Arc Total-Body Irradiation - Held2012
2012 Dosimetric Aspects of Inverse-Planned Modulated-Arc Total-Body Irradiation - Held2012
Key words: total-body irradiation (TBI), treatment techniques, inverse planning, full-body dose
distribution, TBI arc treatment
5263 Med. Phys. 39 (8), August 2012 0094-2405/2012/39(8)/5263/9/$30.00 © 2012 Am. Assoc. Phys. Med. 5263
5264 Held et al.: Dosimetric aspects of modulated-arc TBI 5264
plans have been calculated on patient CT scans. Each plan supine beams in all directions, i.e., 1, 5, and 10 cm in the
was evaluated with respect to dose uniformity and treatment lateral, inferior, and anterior directions. The position of the
time. blocks relative to the patient remained the same. Each time,
During patient simulation, radiopaque markers are placed the dose distribution was recomputed. This offset introduced
on the front and back of the patient, inline with the sternum. an error that corresponds to a patient misalignment between
The beams are aligned with the patient such that the field cen- supine and prone setup during treatment. Despite the 10 cm
ter of the 0◦ gantry angle beam aligns with these setup points. shift, the body contour was still entirely covered by the treat-
For treatment planning, a dosimetrist contours the outline of ment beams. The impact on dose distribution within the body
the body contour and the lungs, which takes about 15 min on is described by the V(±10) for the external body contour. To
average. The physician draws the contours for the lung blocks indicate the impact of an isocenter shift on the dose distribu-
and any additional organ at risk. tion within the lungs, maximum, and mean doses are reported
Prior to the optimization, beams were set up so that the for the left lung contour. The patient plan utilized for this test
most outside ones contain at least 5 cm of the target in the had a prescription of 150 cGy per fraction to a percentage of
beam field at isocenter, which can help reduce the total MU the mean body contour minus the skin, i.e., 5 mm.
of the plan. Furthermore, the fields were distributed along the
arc with an angular offset between supine and prone beams
to avoid parallel-opposed beams. The dose for these beams is II.F. Anthropomorphic water phantoms
calculated using the adaptive convolve algorithm in Pinnacle. Three water phantoms of different sizes were constructed
Thus, body inhomogeneity is corrected for. Inverse-planned for the purpose of dose verification measurements and op-
beam-weight optimization with a single constraint of uni- timizing treatment planning parameters. The phantoms are
form dose to the body contour was used to calculate an opti- made from 0.64 cm thick acrylic sheets that were glued to-
mal dose distribution. This optimization process takes roughly gether to form boxes of different shapes to represent a patient
15 min on up-to-date Pinnacle stations. To investigate the op- body. These three phantoms had the lengths of 71, 113, and
timal field size, the field length covering the craniocaudal di- 170 cm, which represent the height of a baby, a child, and an
rection of the patient was varied between 20, 30, and 40 cm. adult, respectively. The adult phantom has ten footprints, la-
The field width that covers the patient laterally was kept at beled with numbers 1 through 10, for placement of an IBA
the maximum of 40 cm for all plans. Treatment times were linear diode array [shown in Fig. 2(a)]. The array measured
calculated and compared for each field size based on the total the dose at a water depth of 5.0 cm in the arms, 8.1 cm in the
amount of MU. Since there are data that suggest that lower legs, and 10.1 cm in the torso. The other two phantoms use
dose rates reduce pulmonary complications,13–17 the dose rate acrylic blocks as mounts for an IBA cc13 ion chamber [see
for the beams that directly irradiate the lungs was set to Fig. 2(b)]. For the baby and child phantoms, these mounts
50 MU/min. To reduce treatment time, all other beams were allow for 11 and 13 different point dose measurements, re-
increased to a dose rate of 300 MU/min. In the next step, the spectively. These are labeled with the number 1 through 6,
total number of beams was reduced from 45 to 23 by increas- to denote the ion chamber mount, and with an additional let-
ing the gantry step size from 5◦ to 10◦ . ter s, i, l, or r to denote the superior, inferior, left, or right
All treatment plans were evaluated based on the dose-
volume histogram (DVH) for the entire patient body. The pa-
tient used for this beam parameter test was prescribed with the
total dose of 200 cGy to the mean body dose. To quantify
the difference in the overall dose distribution, the fraction
of the volume within ±10% of the prescription dose, V(±10),
was extracted from each plan. The rationale for this number
was that the aim of TBI is to treat the whole body within
±10% of the prescribed dose.5 In addition, the standard de- (a)
viation for the mean patient dose relative to their prescription
dose was extracted from the TPS.
orientation of the ion chamber in the mount. The ion cham- gle beam field. Then, the supine treatment for a patient was
ber was positioned at a water depth of 6.6 cm in the torso of delivered to the phantom and a dose profile was taken with
both phantoms. The dose was measured at 4.1 cm depth in the the linear diode array placed in footprint 4 [see Fig. 2(a)] at
legs of the baby model. There are no individual arms to this 13.8 cm below the block, which corresponded to a water
phantom. Measurement points for the child phantom were at depth of 10.1 cm. Likewise, the blocks were included in the
4.4 cm water depth inside the legs and 4.1 cm inside the arms. TPS at the same position relative to the CT images. After
overriding the lung block density and computing the dose
distribution, the predicted dose profile for the diode array was
II.G. Dose delivery measurements
extracted from the treatment plan and compared to the mea-
Absolute and relative dose measurements were performed sured values.
on the anthropomorphic water phantoms. In preparation, a CT
scan of each water phantom was acquired. The density at the
water surface was adjusted to eliminate the waves caused by II.I. Effect of rotational delivery across blocks
the CT table motion. An MATBI treatment plan was applied The use of gantry rotation to deliver TBI blurs the dose dis-
to each phantom scan and the resulting dose distribution was tribution under the cerrobend blocks, causing an increase in
computed. dose penumbra along the superior/inferior direction. The lung
For the absolute dose point measurement, the child phan- blocks used during treatment are nondivergent, as each beam
tom was placed on the treatment couch. All supine beams irradiates the target from a different gantry angle. The blocks
of the MATBI treatment plan were delivered for each of the are produced such that the partial lung volume is blocked with
13 measurement points. The point dose was measured with a 1 cm margin inside the outline of the lungs. To quantify the
an IBA cc13 ion chamber placed inside the acrylic mounts effect of the blurring, the dose distribution for rotational de-
and was averaged over both negative and positive electrom- livery across physical lung blocks has been studied in com-
eter polarities. Absolute dose point measurements were also parison to that for AP fields, using the ability of modeling
performed for the baby phantom, which was built to fit in- cerrobend blocks in Pinnacle. For this purpose, an MVCBCT
side a single open-field beam. To maximize the field length, image of the blocks was fused with the CT scan images of the
the collimators were rotated by 45◦ and one diamond-shaped adult phantom. The blocks were roughly centered on the 0◦
beam field with 40 × 40 cm2 field size at isocenter was used gantry angle beam field as they are during MATBI treatment.
to deliver the planned dose. This was repeated for seven cc13 As previously described, the block density was overridden
positions throughout the phantom. The dose was again deliv- to 9.3 g/cm3 . Then, the dose distribution was computed for
ered for both polarities. an AP field and for an MATBI treatment, with gantry angles
Relative dose measurements were performed with the adult between 295◦ and 35◦ . Additionally, the same computations
phantom. For each of the ten footprints, the supine beams for were performed in the absence of blocks. The transmission is
an MATBI plan were delivered. The dose profile was mea- computed as the quotient of the dose with and without blocks.
sured with the IBA linear diode array that has 99 diodes, This calculation was performed for the depths of 4, 8, 12, and
spaced 0.5 cm apart. 16 cm below the blocks.
Prediction Measurement
1.1 1 2
0.9
1.1 3 4
0.9
Relative Dose [%]
1.1 5 6
0.9
1.1 7 8
1
F IG . 4. Relative amount of backscattered radiation off the floor measured
0.9 with build-up material around the ion chamber.
1.1 9 10
These show that, when the gantry step size remains constant,
1
dose uniformity is best for the largest field size of 40 × 40
0.9 cm2 . For fields of 40 × 40 cm2 , 30 × 40 cm2 , and 20 × 40
−20 −10 0 10 20 −20 −10 0 10 20 cm2 size, V(±10) equals 89.5%, 87.9%, and 82.5%, respec-
Diode Position [cm]
tively. The corresponding standard deviation of the body dose
F IG . 3. Relative dose comparison for the predicted and measured dose, relative to the mean dose of 200 cGy is 8.7%, 9.0%, and 9.8%,
with the linear diode array for ten different positions inside the adult water respectively. Dose uniformity decreases with the number of
phantom. beams used during MATBI treatment. V(±10) is 89.5% for
45 treatment fields (5◦ step size) as opposed to 79.5% for 23
between −3.2 and 3.7% when using the old beam model for
beams (10◦ step size). The corresponding standard deviation
the dose computation. The new and old beam models pro-
relative to the mean dose is 8.2% and 9.5%, respectively. For
vided an average absolute difference of 1.2% and 1.4%, re-
a prescribed dose of 200 cGy, the estimated beam-on time for
spectively, from the measured values. Dose measurements for
MATBI supine and prone treatment is 67 min for a 190 cm tall
the seven positions inside the baby phantom were within −0.3
person with a 40 × 40 cm2 field size and a 50 MU/min dose
and 1.7% of the dose calculations for the new beam model
rate for all beams. Treatment time increases to 81 min and
and between −1.0 and 2.4% for the old beam model. For this
phantom, the new and old beam models had an average abso-
lute difference of 1.2% and 1.8%, respectively. Figure 3 shows
Volume Above Dose [%]
1
the profile for each diode array position inside the adult phan- 40 cm by 40 cm
30 cm by 40 cm
tom. The measured dose profiles inside the adult water phan- 20 cm by 40 cm
tom show undulating patterns along the array. The mean ab- 0.5
solute difference between the measured and calculated dose
for each of the ten footprints is less than 3%.
0
1.6 1.7 1.8 1.9 2 2.1 2.2 2.3 2.4
Dose [Gy]
III.C. Backscatter
Volume Above Dose [%]
1
Figure 4 shows the relative amount of backscatter mea- 45 beams
23 beams
sured at 5, 10, 20, and 30 cm above the floor with buildup ma-
terials wrapped around an ion chamber. The relative backscat- 0.5
ter decreases with increasing distance to the floor. At 20 cm
above the floor, which is roughly the distance of the treatment
couch top to the floor, the relative backscatter contribution re- 0
duces to 1.6%. 1.6 1.7 1.8 1.9 2 2.1 2.2 2.3 2.4
Dose [Gy]
1 TABLE II. Change of the maximum and mean dose inside the left lung con-
no shift tour for a 1, 5, and 10 cm shift into lateral, craniocaudal, and SSD direction
0.9 10 cm lateral relative to the original plan in percent. For no shift, the maximum dose is
10 cm superior
167 cGy and the mean dose is 116 cGy per fraction inside the left lung
0.8 10 cm shorter SSD
contour.
Volume Above Dose [%]
0.7
Relative dose difference in lungs (%)
0.6
Max Mean
0.5
Setup error (cm) Lateral Craniocaudal SSD Lateral Craniocaudal SSD
0.4
1 −0.13 −0.19 −0.68 −0.02 −0.10 −0.69
0.3 5 −1.50 −0.13 −5.77 −0.33 −0.82 −3.65
0.2 10 −2.34 −0.54 −14.39 −0.70 −0.85 −7.81
0.1
dard deviation relative to the mean dose is 11.9%. Table I lists 0.9
the V(±10) for lateral, craniocaudal and decreased SSD shifts
0.8
TABLE I. Amount of body volume that is within ±10% of the prescribed 0.7
dose when introducing a setup error of 1, 5, and 10 cm to the prone isocenter.
0.6
Body volume inside ±10% of prescribed dose (%)
0 20 40 60 80 100
1 78.55 78.54 75.72 Distance [cm]
5 78.92 78.55 57.15
10 79.88 78.78 25.66 F IG . 7. Predicted and measured dose profile underneath a physical lung
block.
4 cm 8 cm
1.1 1.1
1 1
Transmission [%]
Transmission [%]
0.9 0.9
0.8 0.8
0.7 0.7
0.6 0.6
0.5 0.5
0.4 0.4
10 5 0 5 10 10 5 0 5 10
Length [cm] Length [cm]
12 cm 16 cm
1.1 1.1
1 1
Transmission [%]
Transmission [%]
0.9 0.9
0.8 0.8
0.7 0.7
0.6 0.6
0.5 0.5
0.4 0.4
10 5 0 5 10 10 5 0 5 10
Length [cm] Length [cm]
straight on field arc fields
F IG . 8. Dose profile predictions for transmitted dose at different distances underneath a physical lung block.
of 55% compared to a predicted average relative dose of 57% However, compared to the prescribed dose, which the patient
below the block. will receive during treatment, the reduction in dose to the pa-
tient during the simulation is clinically irrelevant.
A new beam model was created specifically for the
III.G. Effect of rotational delivery across blocks extended-SSD patient position of MATBI to investigate the
Figure 8 displays the calculated dose transmission profile, potential dose accuracy improvements. One challenge in cre-
which is the quotient of the dose inside the phantom with and ating a suitable beam model for MATBI is obtaining suffi-
without blocks, along the superior-inferior direction at differ- cient accuracy in the tail of the dose distribution. Since the
ent depths. Profiles are calculated for distances of 4, 8, 12, and radiation field is swept along the full length of the patient,
16 cm below the block. The angular beam incidence causes each point of the anatomy receives dose from the tail of many
dose to reach into the region underneath the block, thus in- different beams. Thus, small errors in the dose-tail modeling
creasing the penumbra with depth. The penumbra is defined can accumulate to make a large discrepancy. The new beam
here as the distance between the 90 and 60% transmission model focused on this issue. Nevertheless, in this case, the
values. It is measured for the inferior and superior edge of the new MATBI beam model shows no real significant improve-
block and is then averaged. At 4 cm, the field penumbra is ment on the average absolute dose error. However, the benefits
1.6 cm for rotational dose delivery as opposed to 0.6 cm for of developing an MATBI-specific beam model would vary at
AP fields. At 16 cm depth, the corresponding penumbra in- different institutions,11, 18 as the accuracy of a beam model in
creases to 5 cm and 0.8 cm, respectively. the dose tail depends not only on the person commissioning
the model but also the TPS itself.
Another consequence of MATBI treatment compared to
IV. DISCUSSION
isocentric irradiation is the presence of additional backscat-
Although lower scan values reduce the CT image qual- tered dose off the floor. Measurements show that the couch
ity, this has no significant effect on the accuracy of treatment height of 20 cm above the floor reduces the relative amount
planning and dose computation. Thus, the amount of data and of backscatter to 1.6%. This together with the good agree-
the time required for contouring can be reduced for scans us- ment between dose calculations and ion chamber measure-
ing larger slice thickness and lower scan values. Acquiring ments was sufficient evidence to neglect backscatter off the
fewer images per patient also minimizes dose to the patient. floor.
The field size of 40 × 40 cm2 was optimal not only for patient in a comfortable and stable position in standard-sized
treatment time, but also dose homogeneity. This might appear vaults. The CT-based inverse planning approach allows dose
counter intuitive, as one would expect smaller field sizes to distribution optimization and evaluation. The MATBI beam
allow for better dose modulation. However, since the gantry parameters have a large influence on the treatment time and
angle step size was fixed, using large beam fields causes more also affect dose uniformity. Consequently, the application of
beams to overlap at each point. The end of one beam along the optimum beam settings that are presented here is crucial when
patient length does not exactly coincide with the beginning of implementing MATBI. This study also lays out the required
another. This introduces a dose undulation with a relative am- measurements to ensure the accuracy of MATBI dose de-
plitude that is inversely proportional to the number of overlap- livery. Both of these dosimetric aspects are essential for the
ping beams. Thus, the higher number of overlapping beams clinical implementation of MATBI. The technique has al-
with a 40 × 40 cm2 field size reduces this amplitude, and ready been used on 23 patients and is now the standard at
therefore also improves homogeneity. Regions of high dose UCSF.
(>10% above the prescription dose) appear within the arms
and fingers. Since MATBI does not modulate the dose later-
ally, these anatomical structures, which are thin compared to ACKNOWLEDGMENTS
the torso, receive a higher dose than the prescription dose. The This work was supported in part by Siemens.
highest dose points appear at the peak of the dose undulations
in the arms and fingers.
In addition to these dose undulations, the intermittent dose
a) Author to whom correspondence should be addressed. Electronic mail:
delivery for TBI causes the potential problem that malignant
jpouliot@radonc.ucsf.edu; Telephone: (415) 353-7190; Fax: (415) 353-
cells might evade irradiation due to cell circulation. Molloy19 9883.
has investigated this issue in detail, using the field and treat- 1 J. C. Breneman, H. R. Elson, R. Little, M. Lamba, A. E. Foster, and
ment time parameters of tomotherapy-based TBI treatment. B. S. Aron, “A technique for delivery of total body irradiation for bone
With reference to these results, the effect of intermittent dose marrow transplantation in adults and adolescents,” Int. J. Radiat. Oncol.,
Biol., Phys. 18, 1233–1236 (1997).
delivery is even less of an issue for MATBI than it was for the 2 S. V. Harden, D. S. Routsis, A. R. Geater, S. J. Thomas, P. J. Taylor,
results in Molloy. MATBI treatment times are long enough to R. E. Marcus, and M. V. Williams, “Total body irradiation using a modified
reduce the effect of the intermittency. Furthermore, each field standing technique: A single institution 7 year experience,” Br. J. Radiol.
irradiates about 80 cm along the axial length of the patient 74, 1041–1047 (2011).
3 C.-S. Chui, D. P. Fontenla, E. Mullokandov, A. Kapulsky, Y.-C. Lo,
and therefore covers a sufficient part of the body at any given and C.-L. Lo, “Total body irradiation with an arc and a gravity-
time. oriented compensator,” Int. J. Radiat. Oncol., Biol., Phys. 39, 1191–1195
The total-body dose coverage is not easily susceptible to (1997).
4 M. Chrétien, C. Côté, R. Blais, L. Brouard, L. Roy-Lacroix, and
errors from misalignments of the patient in the lateral or
M. Larochelle, “A variable speed translating couch technique for total body
superior-inferior directions. This alignment is, however, more irradiation,” Med. Phys. 27, 1127–1130 (2000).
critical for organs at risk. Maintaining the correct SSD is cru- 5 M. Sarfaraz, C. Yu, D. J. Chen, and L. Der, “A translational couch tech-
cial to the dose for both the total body and the organs at risk. nique for total body irradiation,” J. Appl. Clin. Med. Phys. 2, 201–209
(2001).
The V(±10) itself is not sufficient to reflect errors in the dose 6 A. Hussain, J. E. Villarreal-Barajas, P. Dunscombe, and D. W. Brown,
distribution caused by misalignment. On the contrary, a slight “Aperture modulated, translating bed total body irradiation,” Med. Phys.
increase of the V(±10) values for lateral shifts (see Table I) 38(2), 932–941 (2011).
7 M. Held, “A new modulated-arc inverse-planned total body irradiation
could be misleading. In addition, the standard deviation and
technique,” Diploma thesis, Department of Physics, University of Ham-
an evaluation of the overall DVH are important for a detailed burg, Hamburg, Germany, 2011.
description of dose errors. 8 N. Kirby, “Inverse-planned modulated-arc total-body irradiation,” Med.
The comparison of measurements and dose calculations Phys. 39, 2761–2764 (2012).
9 T. B. Shope, R. M. Gagne, and G. C. Johnson, “A method for describing the
shows that Pinnacle can model the presence of high-density
doses delivered by transmission x-ray computed tomography,” Med. Phys.
materials for lung blocks. Thus, the effect of these blocks 8, 488–495 (1981).
can be included into patient dose calculations. The ability to 10 H. D. Nagel, Radiation Exposure in Computed Tomography (COCIR,
model the OAR blocks was utilized to investigate the effect Frankfurt, 2000).
11 M.-C. Lavallée, L. Gingras, M. Chrétien, S. Aubin, C. Côté, and
of rotational delivery across physical lung blocks. This in-
L. Beaulieu, “Commissioning and evaluation of an extended SSD photon
vestigation demonstrates that compared to AP beams MATBI model for PINNACLE3 : An application to total body irradiation,” Med.
yields similar minimum block transmission values, but that Phys. 36, 3844–3855 (2009).
12 J. Van Dam, A. Rijnders, L. Vanuytsel, and H.-Z. Zhang, “Practical
the dose penumbra increases with depth. For this reason, it is
important to place the blocks as close to the patient as possi- implication of backscatter from outside the patient on the dose distri-
bution during total body irradiation,” Radiother. Oncol. 13, 193–201
ble, ideally onto the patient’s skin. (1988).
13 J. Van Dyk, T. J. Keane, S. Kan, W. D. Rider, and C. H. J. Fryer, “Radiation
15 R. S. Weiner, M. M. Bortin, R. P. Gale, E. Gluckman, H. E. Kay, H. J. Kolb, interstitial pneumonitis after bone marrow transplantation,” Tohoku J. Exp.
A. J. Hartz, and A. A. Rimm, “Interstitial pneumonitis after bone marrow Med. 202, 255–263 (2004).
transplantation,” Ann. Intern. Med. 104(2), 168–175 (1986). 18 A. Hussain, E. Villarreal-Baraja, D. Brown, and P. Dunscombe, “Validation
16 C. A. Barker, T. J. LoSasso, and S. L. Wolden, “Total body irradiation,” of the eclipse AAA algorithm at extended SSD,” J. Appl. Clin. Med. Phys.
Textbook of Radiation Oncology, edited by T. L. Phillips, R. T. Hoppe, and 11, 90–100 (2010).
M. Roach III (W. B. Saunders, Philadelphia, 2004). 19 J. A. Molloy, “Statistical analysis of dose heterogeneity in circulating
17 M. Beyzadeoglu, K. Oysul, B Dirican, F. Arpaci, A Balkan, S. Surenkok, blood: Implications for sequential methods of total body irradiation,” Med.
and Y. Pak, “Effect of dose-rate and lung dose in total body irradiation on Phys. 37, 5568–5578 (2010).