INSTITUTE OF PHYSICS PUBLISHING Phys. Med. Biol. 51 (2006) 2987–2995
PHYSICS IN MEDICINE AND BIOLOGY
doi:10.1088/0031-9155/51/11/020
Use of peripheral dose data from uniform dynamic multileaf collimation fields to estimate out-of-field organ dose in patients treated employing sliding window intensity-modulated radiotherapy Shamurailatpam Dayananda Sharma, Ritu Raj Upreti and Deepak Dattatray Deshpande Department of Medical Physics, Tata Memorial Hospital, Dr Ernest Borg’s Marg, Parel, Mumbai 400 012, India E-mail: shamu
[email protected]
Received 30 January 2006 Published 24 May 2006 Online at stacks.iop.org/PMB/51/2987 Abstract Peripheral doses (PD) from uniform dynamic multileaf collimation (DMLC) fields were measured for 6 MV x-rays on a Varian linear accelerator using a 0.6 cc ionization chamber inserted at 5 cm depth into a 35 × 35 × 105 cm3 plastic water phantom. PD measurements were also carried out under identical conditions for seven patients treated for head and neck and cervical cancer employing sliding window intensity-modulated radiotherapy (IMRT). The measured PD from these patient-specific intensity-modulated beams (IMBs) were compared with the corresponding data from uniform DMLC fields having similar jaws setting. The measured PD per monitor unit (PD/MU) decreases almost exponentially with out-of-field distance for all uniform DMLC and static fields. For the same strip field width of 1.2 cm, uniform DMLC fields with a larger size of 14 × 22 cm2 deliver an average of 3.51 (SD = 0.51) times higher PD/MU at all out-of-field distances compared to 6 × 6 cm2. Similar to uniform DMLC fields, PD/MU measured from different patientspecific IMBs was found to decrease almost exponentially with out-of-field distance and increase with increase in field dimension. PD per MU from uniform DMLC fields and patient-specific IMBs having similar jaws setting shows good agreement (±7%) except at the most proximal distance, where a variation of more than 10% (maximum 15%) was observed. Our study shows that PD data generated from uniform DMLC fields can be used as baseline data to estimate out-of-field critical organ or whole-body dose in patients treated employing sliding window IMRT if an appropriate correction factor for field dimension is applied. The whole-body dose information can be used to estimate
0031-9155/06/112987+09$30.00 © 2006 IOP Publishing Ltd Printed in the UK
2987
2988
S D Sharma et al
the possible increase in risk of fatal secondary malignancy in patients treated employing sliding window IMRT. (Some figures in this article are in colour only in the electronic version)
1. Introduction Intensity-modulated radiotherapy (IMRT) is widely adopted in the treatment of several types of cancer because of its superior dose conformity to irregular target volume and conformal avoidance of nearby critical organs. However, this increasingly popular technique requires more monitor units (MUs) per target dose, resulting in higher secondary dose outside the treatment field (peripheral dose), as compared to conventional radiotherapy (Followill et al 1997, Kry et al 2005a, Meeks et al 2002, Mutic and Low 1998, Vanhavere et al 2004). Though peripheral dose (PD) is an unavoidable consequence of any type of radiation therapy, higher PD reported from IMRT may increase the risk of radiation-induced secondary malignancies, particularly in patients with long-term survivors (Followill et al 1997, Kry et al 2005b, Meeks et al 2002, Mutic and Low 1998, Vanhavere et al 2004). Increase in PD from IMRT is widely variable depending on the intensity-modulated beam (IMB) delivery technique, complexity of IMB, linear accelerator head design, treatment site and x-ray energy. Very limited information is available on the PD characteristic from IMRT. Vanhavere et al (2004) reported peripheral photon and neutron dose measurement for a single prostate treatment using sliding window IMRT. In view of the increasing application of IMRT in varieties of clinical sites, baseline PD data need to be established for each technique to estimate any out-of-field organ dose. In our previous work, increase in PD has been reported from uniform dynamic multileaf collimation (DMLC) fields compared to static open fields (Sharma et al 2006). In this study, PD data measurement from uniform DMLC fields was extended for more field sizes to account for the spectrum of clinical sites. To evaluate the applicability of these data to actual patient treatments employing sliding window IMRT, PD was also measured for different clinically planned intensity-modulated beams having different jaws settings and beamlet patterns varying in complexity. 2. Materials and method Static fields of 6 × 6, 10 × 10, 14 × 14, 14 × 18 and 14 × 22 cm2 were simulated in uniform DMLC mode by moving strip fields of 1.2 cm width created by 26 pairs of MLCs, each projecting a leaf width of 1 cm at isocentre. Additional strip fields having constant widths of 0.5, 1, 1.5 and 2 cm were also used to simulate 14 × 14 cm2. Peripheral dose from these uniform DMLC fields was measured for 6 MV x-rays on a Clinac 2100 C/D linear accelerator (Varian Associates, Palo Alto, CA) using a 0.6 cc Farmer type ionization chamber (PTW Friedberg, Germany) inserted at 5 cm depth into a 35 × 35 × 105 cm3 plastic water phantom (Nuclear Associate, USA) under isocentric conditions. Figure 1 shows the schematic diagram of the collimator design of the Clinac 2100 C/D linear accelerator and experimental set-up for the measurement. The details of PD measurements from uniform DMLC fields have been described in our previous work (Sharma et al 2006). The peripheral doses per monitor unit (PD/MU) measured at different out-of-field distances from these DMLC fields were compared with corresponding data from clinically used IMBs of different complexities. Seven patients treated for head and neck and cervical cancer (3 nasopharynx, 1 ethmoid sinus and 3 cervical cancers) employing sliding window IMRT were selected for this study.
2989 MLC motion direction
Extra target dose from sliding window IMRT
0.0 cm 9.5 cm 12.4 cm 16.91 cm
Measurement direction
Y 34.97 cm
X
44.44 cm
MLC 53.50 cm
5 cm O
*
* *
*
*
*
*
* *
*
0.6 cc ion chamber O
* - Represent measurement points; not scaled
Figure 1. Schematic diagram of the collimator design of the Clinac 2100 C/D linear accelerator and experimental set-up for the PD measurement from uniform DMLC fields and patient-specific IMBs.
As a part of an ongoing randomized trial, patients of stage IIB cervical cancer are treated with a dose of 50 Gy in 25 fractions using sliding window IMRT and 35 Gy in 5 fractions using intracavitory brachytherapy. IMRT plans of head and neck patients selected for this study are taken from two separate trials in adult and paediatric patients, treated with a dose of 70 Gy. All these patients were selected to account for different jaws settings, MUs and complexities of IMBs. All patients were planned and treated as per the standard IMRT protocol of the institute. Optimization, dose computation and leaf sequencing were performed on a Cadplan version 6.2.7 (Varian Associates, Palo Alto, CA) treatment planning system (TPS). Five to nine co-planer fields were employed for each patient. In patients with a large tumour, wherein field size bigger than 14.5 cm is employed in radial directions (X), intensity-modulated fields were automatically split into two parts using the dose feathering methods. A dose of 2 Gy/Fr was planned and delivered to the same linear accelerator (Clinac 2100 C/D) using 6 MV x-rays. The IMB’s fluence of these patients was used for the measurement of PD in the phantom. For this purpose, all the IMBs of the final plan of every patient were transferred to a water phantom in the same TPS at gantry zero and target-to-surface distance (TSD) of 95 cm. For each plan, total dose at isocentre, which is at 5 cm depth, was computed and adjusted to 2 Gy. The IMBs along with the corresponding MUs were transferred to the Clinac 2100 C/D linear accelerator and the isocentre dose was verified by ion chamber measurements. Then PD data from these IMBs were also measured at different out-of-field distances in the same way as done for uniform DMLC fields (Sharma et al 2006). 3. Results The number of MUs required to deliver 2 Gy at 5 cm depth from the uniform DMLC field of 14 × 14 cm2 increases from 1102 for 2 cm strip field width to 3052 for 0.5 cm strip field width. Whereas to deliver the same dose keeping a constant strip field width of 1.2 cm, a larger field of 14 × 22 cm2 required 1766 MUs compared to 844 MUs for a smaller field of 6 × 6 cm2.
2990
S D Sharma et al 0.1
MLC Static field 0.5 cm strip field width 1 cm strip field width
PD/MU (cGy/MU)
1.5 cm strip field width 2 cm strip field width
0 .0 1
0.001
0 .0 00 1 0
10
20
30
40
50
60
70
80
Distance from the central axis (cm) Figure 2. Peripheral dose per MU (PD/MU) expressed as cGy/MU at 5 cm depth and 14 × 14 cm2 fields simulated in uniform DMLC mode using different strip field widths.
In all patients, the treatment planning system calculated and measured isocentre dose at 5 cm depth agrees within ±2.3% (range 1.8–3.4%). Figure 2 shows the peripheral dose per MU expressed in cGy/MU, measured at 5 cm depth and various out-of-field distances from the central axis of 14 × 14 cm2 fields simulated in uniform DMLC mode using different strip field widths. The measured PD/MU decreases almost exponentially with out-of-field distance for all DMLC and static fields. However, within the out-of-field distance of 40 cm from the beam central axis, PD/MU was found to deviate slightly from its exponential nature forming crests and troughs of slightly varying amplitude. All DMLC fields deliver almost the same PD/MU at all out-of-field distances except in the proximal region wherein the largest strip field width of 2 cm delivers a maximum of 1.5 times higher dose compared to the smallest strip field width of 0.5 cm. Figure 2 also reveals that PD/MU from the static open field shaped by stationary MLCs was almost four times higher than that of DMLC fields at the most proximal distance of 12 cm from the field centre, which progressively decreases to 1.09 at 67 cm. Peripheral dose per MU was also found to vary with field size. For the same strip field width of 1.2 cm, uniform DMLC fields with larger size were found to deliver higher PD/MU at all out-offield distances as shown in figure 3. DMLC fields of 14 × 22 cm2 deliver an average of 3.51 (SD = 0.51) times higher PD/MU compared to 6 × 6 cm2 at all out-of-field distances. While for the same jaws width of 14 cm, DMLC with 22 cm in length delivers an average of 1.49 (SD = 0.27) times higher PD/MU than that of 14 cm in length. Table 1 shows the average jaws setting of all IMBs for each patient’s plan and corresponding MU required for delivering 2 Gy at 5 cm depth in a water phantom at zero gantry position. For each plan, the Y-jaws dimension is almost the same for all IMBs, while the X-jaws dimension varies from field to field. The jaws define average field dimension for the different patients’ plans ranging from 11.8 × 13.4 cm2 to 15 × 20.5 cm2. In four patients, where the field width is larger than 14.5 cm, each field was split into two parts due to the limitation of the Varian MLC. The number of MUs required for delivering 2 Gy at 5 cm depth from the different patient-specific IMBs varied from 527 to 972 for head and neck and from 632 to 1244 for cervical cancer patients. Similar to uniform DMLC fields, PD/MU
Extra target dose from sliding window IMRT
2991
0 .1
DMLC_6x6 cm2 DMLC_10x10 cm2 DMLC_14x14 cm2 DMLC_14x18 cm2
PD/MU (cGy/MU)
DMLC_14x22 cm2 0.01
0.0 01
0.00 01 0
10
20
30
40
50
60
70
80
Distance from the central axis (cm) Figure 3. Peripheral dose per MU (PD/MU) expressed as cGy/MU at 5 cm depth and different field sizes simulated in uniform DMLC mode using the same strip field width of 1.2 cm. Table 1. Treatment planning parameters of the seven patients treated employing sliding window IMRT.
Patient
Diagnosis
No of fields
P1 P2 P3 P4 P5 P6 P7
Ca. Ca. Ca. Ca. Ca. Ca. Ca.
7 9 5 × 2 = 10 5 × 2 = 10 7 5 × 2 = 10 5 × 2 = 10
nasopharynx nasopharynx cervix cervix ethmoid sinus cervix nasopharynx
Average jaws setting (X × Y cm2)
MU required to deliver 2 Gy at 5 cm depth
11.8 × 13.4 12 × 13.5 19.5 × 16.4 18.5 × 16.8 12.6 × 17.7 15 × 18.5 15 × 20.5
730 752 632 740 527 1244 972
measured at 5 cm depth from different patient-specific IMBs was also found to decrease almost exponentially with out-of-field distance from the beam central axis and increase with increase in field dimension (figure 4). PD per MU data from uniform DMLC fields having 1.2 cm strip field width are interpolated to yield a PD/MU value for field dimension (specially Y jaws) similar to that of patient-specific IMBs. Table 2 represents the percentage variation between interpolated PD/MU from uniform DMLC fields and patient-specific data at different outof-field distances. PD per MU from uniform DMLC fields and patient-specific IMBs having similar jaws settings agree within ±7% for all patients except at the most proximal distance of 17 cm from the field centre where a variation of more than 10% (maximum 15%) was observed. 4. Discussion While IMRT reduces normal tissue complication probability, it also increases whole-body dose due to the increase in overall MUs. This may subsequently lead to higher risk of radiation-
2992
S D Sharma et al
PD/MU (cGy/MU
0.1
P1_12x13.4 cm2 P2_12x13.5 cm2 P3_19.5x16.4 cm2 P4_18.5x16.8 cm2 P5_12.6x17.7 cm2 P6_15x18.5 cm2 P7_15x20.5 cm2
0.01
0.001
0.0001 0
10
20
30
40
50
60
70
80
Distance from the central axis (cm) Figure 4. Peripheral dose per MU (PD/MU) expressed as cGy/MU measured at 5 cm depth from different patient-specific IMBs.
induced secondary malignancies, particularly in patients with long-term survivors (Followill et al 1997, Hall and Wuu 2003, Kry et al 2005b, Mutic and Low 1998). Followill et al (1997) reported theoretically estimated whole-body equivalent dose from step-and-shoot and serial tomotherapy IMRT. Their estimation is based on the calculation of PD at 50 cm from the centre of a 20 × 20 cm2 open field using data previously measured by Stovall et al (1995). They arrived at three and eight times increase in the whole-body dose, a factor similar to the corresponding increase in MUs from step-and-shoot and serial tomotherapy IMRT compared to the conventional technique. For a target dose of 70 Gy, Mutic and Low (1998) reported whole-body equivalent dose measured at the midplane of the patient treated using serial tomotherapy from 6 MV x-rays (Clinac 6/100, Varian Corp.) to be approximately 300 mSv as opposed to the 543 mSv predicted by Followill et al (1997). Similar to the finding of Followill et al (1997), Verellen and Vanhavere (1999) also reported 8.1 times increase in whole-body dose from serial tomotherapy of head and neck patients treated on KDS-2 Mevatron (Siemens, CA) compared to the conventional technique. However, Verellen and Vanhavere observed in vivo measured whole-body effective dose of 1969 mSv, which was much higher than the data reported by other investigators. The corresponding estimated percentage likelihood of a fatal secondary cancer reported by several groups varies from 1.35% (Mutic), 2.7% (Stovall) and 9.9% (Verellen). For a specific prostate cancer treatment with a dose of 70 Gy using 18 MV x-rays and the sliding window IMRT technique, Vanhavere et al (2004) reported an effective dose of around 1000 mSv and associated lifetime risk for an excess fatal cancer of 5%. The difference in whole body and hence risk of radiation-induced secondary malignancies is thus primarily due to the variation in the dose per MU, beam energy and machine collimator design. Moreover, dose per MU varies with disease site, complexity of the IMBs and its delivery technique. Therefore, whole-body dose measured for a particular site may not be applicable to other clinical sites, even for the same delivery technique and treatment machine. Also, most of the available whole-body data are measured at a specific distance (∼50 cm) from the field centre and do not address the dose to critical organs at other distances. Moreover, it may not be applicable to paediatric patients in which the reference point of measurement is much less than 50 cm.
Uniform DMLC Patient-specific IMB
13.5 × 13.5 P1, 12 × 13.4
Distance from field centre (cm) 17 27 37 47 57 67
13.5 × 13.5 P2, 12 × 13.5
14 × 16.4 P3, 19.5 × 16.4
14 × 16.8 P4, 18.5 × 16.8
14 × 17.7 P5, 12.6 × 17.7
14 × 18.5 P6, 15 × 18.5
14 × 20.5 P7, 15 × 20.5
Extra target dose from sliding window IMRT
Table 2. Percentage variation between interpolated PD/MU from uniform DMLC fields and patient-specific data at different out-of-field distances.
Percentage variation between cGy/MU of uniform DMLC fields and corresponding data from patient-specific IMB having similar average jaws setting (cm2) 6.18 −1.06 −4.91 −5.45 6.11
−0.39 −6.05 −6.99 −7.07 −3.22
14.6 5.17 0.34 6.02 5.09 4.25
12.84 6.67 1.14 1.83 6.69 4.62
12.34 3.46 0.08 0.00 3.76 1.07
6.08 5.68 −6.33 6.81 6.38 5.45
12.22 5.36 3.89 5.40 5.38 4.70
2993
2994
S D Sharma et al
Vanhavere et al (2004) and very recently Kry et al (2005a, 2005b) have reported in a phantom measured PD data at different organs and associated risk of fatal secondary malignancies from prostate treatment for sliding window and step-and-shoot IMRT. However, data presented are only for an average jaws setting ranging from 90 cm2 to 100 cm2 and this may not be true for other sites where different jaws openings and various levels of intensity modulations are employed. Kry et al (2005a) also reported higher whole-body dose from the Siemens accelerator as compared to the Varian. It is clear from the above discussion that PD must be measured for the techniques to be used on the treatment machine. To our knowledge, no baseline data are available which can be used to measure PD at various out-of-field distances from sliding window IMRT. In the present study, measured PD from uniform DMLC fields is compared with that from patient-specific IMB generated using the sliding window technique. Patients are selected to account for various complexity levels of IMB and jaws settings commonly encountered in clinical situations. In figure 2, PD/MU of the MLC-shaped static field is almost four times larger than that of the uniform DMLC field at the most proximal distance of 12 cm from the field centre. This is due to the increase in collimator scatter and transmission from the open field which becomes less significant at a larger distance. At larger distances from the field centre, leakage becomes the predominant component of PD and as a result PD/MU from uniform DMLC and open fields are comparable. This explanation holds true for uniform DMLC fields having different sweeping gap widths. Similar to our finding, Kry et al (2005a) also reported approximately two times higher PD/MU at distances up to 20 cm from the central axis from the conventional treatment as compared to step-and-shoot IMRT and this factor reduces to almost one at larger distances. The measured PD/MU at a fixed distance from the field centre increases with the increase in field dimension. This is because of the fact that as field size increases MU/cGy increases and the measurement point becomes closer to the field edge. The maximum field width (X) in uniform DMLC fields was restricted to 14 cm due to the limitation of the Varian MLC, in which the maximum trailing and leading positions of two consecutive MLCs is 14.5 cm. Sharma et al (2006) suggested that the crest and trough pattern of varying amplitudes appearing within 40 cm from the beam central axis in figures 2 and 3 may be due to the position of the MLC with respect to the collimators in the treatment head, MLC design and length above other factors of internal scatter and leakage. Nevertheless, a detailed study needs to be carried out to explain the different contributing factors. The uncertainties associated with low dose measurement especially at far distances are not considered in the presented data. However, the reproducibility of meter readings tested for a few DMLC fields at all out-of-field distances was observed to be within ±2%. Whereas for the static open field and at out-of-field distance more than 47 cm from the central axis, MUs were scaled by a factor of 2 to get good meter readings reproducible within ±2%. The similarity in PD characteristic (PD/MU) observed for uniform DMLC fields and clinically used IMBs demonstrates the applicability of the former data to any clinical situation treated using sliding window IMRT if appropriate corrections for field size are applied. As the conventional method of calculating equivalent field size may not be applicable in the case of IMRT, emphasis has been given to matching specially the Y-jaws setting of uniform DMLC fields and patient-specific IMBs while interpolating data for uniform DMLC fields. Therefore, X-jaws settings between interpolated uniform DMLC fields and patient-specific IMBs differ in all patients. The case of ethmoid sinus (P1) represents simple IMBs, while nasopharynx and cancer of the cervix represent moderate-to-complex IMB fluences. In four patients (P3, P4, P6 and P7), each field was split into two parts as the required field width was more than 14.5 cm. The percentage variation in PD/MU of uniform DMLC fields and corresponding data from patient-specific IMB having similar (Y) jaws setting (cm2) was maximum (15%) at the most
Extra target dose from sliding window IMRT
2995
proximal distance of 12 cm from the field centre. This could be due to the larger contribution of collimator scatter, transmission and internal scattering of complex IMBs compared to uniform DMLC fields at the proximal distances. 5. Conclusion Our study shows that the peripheral dose data generated from uniform DMLC fields can be used as baseline data to estimate out-of-field critical organ or whole-body dose in patients for any clinical sites treated employing sliding window IMRT if an appropriate correction factor for field dimension is applied. The knowledge of whole-body dose can be used to estimate the increase in risk of fatal secondary malignancy from this increasingly popular technique of radiation therapy. This factor should be considered when choosing the optimal treatment technique and delivery system especially in patients with expected long-term survivors. References Followill D, Geis P and Boyer A 1997 Estimates of whole-body equivalent dose produced by beam intensity modulated conformal therapy Int. J. Radiat. Oncol. Biol. Phys. 38 667–72 Hall E J and Wuu S 2003 Radiation-induced second cancers: the impact of 3DCRT and IMRT Int. J. Radiat. Oncol. Biol. Phys. 56 83–8 Kry S, Salehpour M, Followill D, Stovall M, Kuban D, White R and Rosen I 2005a Out-of-field photon and neutron dose equivalents from step-and-shoot intensity-modulated radiation therapy Int. J. Radiat. Oncol. Biol. Phys. 62 1204–16 Kry S, Salehpour M, Followill D, Stovall M, Kuban D, White R and Rosen I 2005b The calculated risk of fatal secondary malignancies from intensity-modulated radiation therapy Int. J. Radiat. Oncol. Biol. Phys. 62 1195–203 Meeks S L, Paulino A C, Pennington E C, Simon J H, Skwarchuk M W and Buatti J M 2002 In vivo determination of extra-target doses received from serial tomotherapy Radiother. Oncol. 63 217–22 Mutic S and Low D A 1998 Whole-body dose from tomotherapy delivery Int. J. Radiat. Oncol. Biol. Phys. 42 229–32 Sharma D S, Animesh, Deshpande S S, Phurailatpam R D, Deshpande D D, Shrivastava S K and Dinshaw K A 2006 Peripheral dose from uniform dynamic multileaf collimation fields: implications for sliding window intensity-modulated radiotherapy Br. J. Radiol. 79 331–5 Stovall M et al 1995 Fetal dose from radiotherapy with photon beams: Report of AAPM Radiation Therapy Task Group No. 36 Med. Phys. 22 63–82 Vanhavere F, Huyskens D and Struelens L 2004 Peripheral neutron and gamma doses in radiotherapy with an 18 MV linear accelerator Radiat. Prot. Dosim. 110 607–12 Verellen D and Vanhavere F 1999 Risk assessment of radiation-induced malignancies based on whole-body equivalent dose estimates for IMRT treatment in the head and neck region Radiother. Oncol. 53 199–203
