A Fong MB BS; R Bromley BMedPhys; M Beat BAppSci; D Vien BAppSci; J Dineley PhD; G Morgan FRCP(UK), FRANZCR.
Presented in part at the Annual Scientific Meeting, Faculty of Radiation Oncology, Royal Australian and New Zealand College of Radiologists, Singapore, November 2006.
Conflicts of interest: None.
Associate Professor Graeme Morgan, Department of Radiation Oncology, Royal North Shore Hospital, Sydney, NSW 2065 Australia. Email: firstname.lastname@example.org
Prior to introducing intensity modulated radiotherapy (IMRT) for whole breast radiotherapy (WBRT) into our department we undertook a comparison of the dose parameters of several IMRT techniques and standard wedged tangents (SWT). Our aim was to improve the dose distribution to the breast and to decrease the dose to organs at risk (OAR): heart, lung and contralateral breast (Contra Br). Treatment plans for 20 women (10 right-sided and 10 left-sided) previously treated with SWT for WBRT were used to compare (a) SWT; (b) electronic compensators IMRT (E-IMRT); (c) tangential beam IMRT (T-IMRT); (d) coplanar multi-field IMRT (CP-IMRT); and (e) non-coplanar multi-field IMRT (NCP-IMRT). Plans for the breast were compared for (i) dose homogeneity (DH); (ii) conformity index (CI); (iii) mean dose; (iv) maximum dose; (v) minimum dose; and dose to OAR were calculated (vi) heart; (vii) lung and (viii) Contra Br. Compared with SWT, all plans except CP-IMRT gave improvement in at least two of the seven parameters evaluated. T-IMRT and NCP-IMRT resulted in significant improvement in all parameters except DH and both gave significant reduction in doses to OAR. As on initial evaluation NCP-IMRT is likely to be too time consuming to introduce on a large scale, T-IMRT is the preferred technique for WBRT for use in our department.
Adjuvant whole breast radiotherapy (WBRT) is an essential component of the management of women with early breast cancer treated by conservative surgery.1 WBRT has produced a reduction in relapse in the treated breast and an improvement in overall survival equivalent to that achieved with adjuvant chemotherapy.
In most centres worldwide the WBRT technique is standard wedged tangents (SWT): two opposed wedged tangential photon beams. However, this results in uneven dose distributions that potentially contribute to late toxicities and poor cosmetic outcomes, such as breast pain and cosmetic outcome.2,3 In addition, the use of SWT that covers all of the ipsilateral breast tissue results in dose to organs at risk (OAR): heart, lung and contralateral breast (Contra Br). Therefore, improving the dose distribution to decrease the ‘hot spots’ and the dose to OAR would be desirable to try and reduce these side effects of WBRT.
Variations in dose distribution in WBRT is a consequence of a number of factors, including the surface contour of the breast, variations in separation supero-inferiorly or medio-laterally, as well as the presence of tissues with varying electron density, such as lung and bone.4 The dose distribution of SWT worsens with increasing breast size, and variations as large as 15–20% have been reported in several studies.5,6 This may also be particularly exaggerated when hypofractionated regimens are utilized.7
Customized physical compensators have been used to improve dose distribution within the breast, but their production is resource and labour intensive.8 The use of static wedges in the linac head is more commonplace, but the dosimetric outcomes are inferior to those obtained with customized compensators and dynamic wedges.
Compared with SWT, the use of intensity modulated radiotherapy (IMRT) for WBRT has been reported by several groups to improve dose distribution and reduce dose to OAR.9–12 As late cardiac and lung toxicity from unnecessary irradiation are the main contributors to long-term morbidity and mortality after breast radiotherapy, any reduction in dose to OAR will lessen the impact of WBRT on long-term survival.
Most of the published data compare a single form of IMRT with SWT and little work has been published comparing several IMRT techniques with SWT. With a view to introducing IMRT for WBRT in our department we compared several methods of IMRT with SWT to determine the relative benefits of each technique. Such a technique would also add little extra workload in a department in which at least 20% of treatments are WBRT.
The study group comprised 20 women who had undergone adjuvant WBRT using 6 MV photons from a Varian linac during 2005 at the Royal North Shore Hospital, Sydney, New South Wales, Australia. The group was selected from our database using a simple computerized randomization and were stratified prior to randomization, so that 10 left-sided breasts and 10 right-sided breasts were evaluated, including small to large breasted women. Although these patients were not specifically selected for anatomy unfavourable for doses to OAR (highly concave breast shapes), several such patients were included in the study group to represent a broad cross-section of women.13 There was no statistically significant difference in breast volume or separation between the two groups (Table 1).
Table 1. Patient characteristics in dosimetric comparison of intensity modulated radiotherapy techniques and standard wedged tangents for whole breast radiotherapy
Mean breast volume (P = 0.47)
Mean breast separation (P = 0.81)
At simulation for SWT the patient was placed in the supine position supported on a Sinmed Posiboard breast board (Sinmed BV, Reeuwijk, the Netherlands) with both arms raised above the head. The treating physician placed opaque flexible markers around the palpable extent of the breast tissue. The treatment field margins were at least 2 cm from the palpated breast tissue and were approximately at the midline for the medial field edge, at the mid-axillary line for the lateral field edge, the level of the sternal notch for the superior field edge and the inferior field edge 2 cm below the infra-mammary fold. CT scans were undertaken using a GE Lightspeed RT 80 cm scanner (GE Healthcare, Rydalmere, NSW, Australia) at 2.5 mm intervals to encompass both breasts and the whole thoracic cavity to include the heart and both lungs. The data were transferred to an Eclipse 3-D treatment planning workstation (Varian Medical Systems, Palo Alto, CA, USA). The CT scan data for the 20 treated patients were retrieved and the original data for breast and the OAR were used for all subsequent plans.
Ipsilateral breast volume
For all plans the original dataset was used to outline the breast: (a) clinical target volume (CTV); (b) planning target volume (PTV); and OAR (c) right lung; (d) left lung; (e) heart; and (f) Contra Br. The CTV was defined as all glandular breast tissue including fatty degenerated ducts of the breast. Delineation of CTV used a fixed window of 0 HU, and a width of 500 HU.14 The CTV was assumed to begin 5 mm below the skin surface, and 5 mm anterior to the lung–chest wall interface.15–17 The PTV was defined as the CTV plus an expansion of 10 mm in all directions except for the skin surface and lung interface. These margins were considered adequate enough to take into account variability in set-up and delineation of CTV.18 The PTV was not expanded to the skin to avoid tissues associated with the optimization algorithm compensating for the build-up region of the depth dose curve19 (Fig. 1). During the planning process the fluence patterns for the electronic compensators IMRT (E-IMRT), tangential beam IMRT (T-IMRT), coplanar multi-field IMRT (CP-IMRT) and non-coplanar multi-field IMRT (NCP-IMRT) were extended to ensure adequate treatment of the PTV volume.
Contralateral breast volume
The Contra Br. tissue was defined as the breast tissue encompassed by the tangential line between the patient’s midline and the contralateral posterior border was defined as being at the same level as the treated breast. The length was defined using the same superior and inferior borders of the treated breast. The contralateral lung was subtracted from the volume. The dose delivered to the Contra Br. was calculated as the mean dose to this volume.20
The heart volume was defined as all the visible myocardium, excluding the pericardium, from the apex of the heart, to the right auricle, atrium, and infundibulum of the ventricle. The ascending aorta, pulmonary trunk, and superior vena cava were excluded.21 The dose to the heart was calculated as the volume receiving 95% of the prescribed dose.22
The lung volume was automatically generated using the auto-contouring tool of the treatment planning system. The irradiated lung volume was expressed as the mean lung dose.23
For SWT the isocentre was midway between the line drawn between the medial and lateral borders in the mid-breast slice in the superior-inferior direction as set by the radiation oncologist at simulation. For IMRT plans the isocentre and gantry angles were adjusted to ensure the PTV was inside the field when viewed in Beams-Eye View (BEV) while avoiding as much heart and lung as possible.
Dose distribution within the ipsilateral breast was assessed by (a) dose homogeneity (DH); (b) conformity index (CI); (c) mean dose; (d) maximum dose; and (e) minimum dose. For DH and CI this required calculating the V 95-107; that is, the volume of the PTV within the 95–107% isodose lines, adhering to International Commission on Radiation Units (ICRU) dosimetric guidelines. DH describes how closely the PTV falls between the 95–107% isodose lines, whereas the CI is the quotient of the treated volume and the PTV. Mean, maximum and minimum doses were derived from point doses. For OAR, the mean lung dose, and mean dose to the Contra Br. were calculated. As heart constraints have not been well established in the published literature to date, we evaluated heart constraints utilizing the cardiac volume receiving 5 Gy (V5), 30 Gy (V30), and 47.5 Gy (V47.5, which is 95% of a standard prescribed whole breast dose).
Planning optimization constraints
For IMRT plans, inverse planning optimization constraints were (a) ipsilateral lung ≤10% should receive 30% of the prescribed dose; (b) contralateral lung ≤10% should receive 5% of the prescribed dose; (c) heart ≤5% should receive 95% of the prescribed dose; and (d) Contra Br. ≤5% to any point of the contralateral breast. These optimization constraints are utilized by Helios (Varian Medical Systems) in the inverse planning of beam fluencies. We elected to utilize these constraints as they conform to the TROG APBI 06-02 protocol for Accelerated Partial Breast Irradiation, and normal tissue tolerances for breast irradiation have not yet been well established.24
All plans were developed using an Eclipse planning system for a Varian linac with dynamic MLC (Varian Medical Systems). The Helios software package was used for inverse planning of IMRT plans.
Standard wedged tangents (SWT)
Radiation fields for SWT were tangential to the chest wall from the lateral to medial markers (as defined) to include all of the PTV. A half-beam block technique on the posterior borders of the medial and lateral fields was used to prevent beam divergence into the lung. The gantry angles were chosen so that the fields were coplanar.
Electronic compensators (E-IMRT)
The tangential fields used for SWT plans were used for E-IMRT, with an irregular surface compensator applied to each field. The irregular surface compensator uses a curved compensation surface (rather than a straight compensation plane) with the compensation point for each beamlet being calculated from the penetration depth in the breast tissue at that level. This depth was defined as a percentage of the distance between the entry and exit point along the beamlet line. A penetration depth of 50% was used for all E-IMRT plans. The beamlet intensity for each ray line should therefore produce an even dose at 50% penetration depth. This method of compensation takes into account the large changes in breast contour and minimizes hot and cold spots in three dimensions. The compensator was edited manually to remove regions of high fluence near the lung–chest wall interface to reduce hot spots in the plan.
Tangential beam IMRT (T-IMRT)
The tangential fields used for SWT plans were used for T-IMRT planning. Dose optimization constraints for the PTV, contralateral breast, heart and lungs were defined as above.
Using an inverse planning algorithm (Helios) a class solution was devised for right- and left-sided tumours. The appropriate class solution was applied and followed by the optimization algorithm. Any unacceptably high doses were reduced by manually editing regions of high fluence that coincided with regions of high dose in each field.
A skin flash was added to each field for all plans in this study using a dynamic MLC. The fluence patterns were edited by manually extending the fluence levels from 2 mm inside the PTV to approximately 2 cm beyond the skin surface. This maintained dose coverage to the PTV while allowing for respiratory motion and variability in treatment set-up.
Multi-field IMRT (CP-IMRT and NCP-IMRT)
A number of different beams and beam arrangements for coplanar and non-coplanar IMRT treatments were produced for seven women. The four-field IMRT beam arrangement that gave the best fit for the PTV and the lowest OAR doses was then adopted for CP-IMRT and NCP-IMRT in all 20 women (Figs 2,3). The same inverse planning class solution process used for T-IMRT was applied for the multi-field techniques. After one optimization, unacceptably high doses were reduced by manually editing regions of high fluence in each field. For NCP-IMRT plans a skin flash was added for all fields as previously defined. For CP-IMRT a skin flash was only added for tangential fields.
The values and ranges of DH, CI and the mean, maximum and minimum breast dose and the OAR (heart, lung and Contra Br.) for the IMRT plans were calculated and compared with the values and ranges of the same parameters for SWT.
For each of the parameters the individual patient results of the IMRT plans were compared with SWT. A Wilcoxon signed rank test was employed using SPSS v16 (SPSS, Chicago, IL, USA).
The mean value and range obtained for each outcome measure and for each treatment technique is presented in Table 2.
Table 2. Outcomes of standard wedged tangents (SWT) compared to intensity modulated radiotherapy (IMRT) plans
The homogeneity index (V95-107) was significantly improved with only the E-IMRT (0.957 vs SWT 0.929, P = 0.003). CP-IMRT resulted in a worse homogeneity index (0.895, P = 0.04). There was no significant difference between the mean doses in the other plans.
Conformity index (CI)
The conformity index was significantly improved with the T-IMRT plans (1.28 vs SWT 1.17, P = 0.04), and the NCP-IMRT plans (1.103, P = 0.001).
The maximum dose was significantly reduced with the E-IMRT plans (108.5% vs SWT 110.2%, P = 0.018), T-IMRT tangents (108.1%, P = 0.02), and with the NCP-IMRT plans (105.9%, P = 0.00001). There was no correlation between breast volume, breast separations, and reductions in maximum dose. (Fig. 4).
The mean dose was lower for the E-IMRT plans (99.3% vs SWT 100.6%, P = 0.004), and CP-IMRT plans (99.6%, P = 0.02), and not significantly different for the other techniques.
There was a significant improvement in dose minimums with the CP-IMRT plans (69.6% vs SWT 49.5%, P = 0.003) and the NCP-IMRT plans (61.8%, P = 0.023).
Organs at risk
The volume of heart receiving large doses (V47.5) was significantly decreased with the T-IMRT (0.56% vs SWT 1.7%, P = 0.011), CP-IMRT (0.24%, P = 0.008), and NCP-IMRT plans (0.25%, P = 0.008). The volume receiving moderate doses (V30) was significantly decreased with the CP-IMRT plans (2.09% vs SWT 4.76%, P = 0.03), and approached significance with the NCP-IMRT plans (2.22%, P = 0.051). The CP-IMRT technique resulted a significantly larger volume of heart receiving very low doses (V5, 27.18% vs SWT 10.68%), but there was no significant difference between all other techniques compared with SWT. Right-sided results were not reported on because no significant cardiac dose was noted with any technique.
Mean lung dose was also significantly reduced with the T-IMRT plans (7.8 Gy vs SWT 9.9 Gy, P = 0.000003), and NCP-IMRT plans (8.2 Gy, P = 0.0001).
Mean dose to the Contra Br. was significantly reduced with the T-IMRT plans (1.8 Gy vs SWT 2.3 Gy, P = 0.01), and the NCP-IMRT plans (1.4 Gy, P = 0.001).
CP-IMRT plans resulted in significantly higher doses to the lung (12.9 Gy, P = 0.000006) and doses to the Contra Br. (6.1 Gy, P = 0.000001).
As a prelude to introducing IMRT for WBRT into our department we undertook the present study to evaluate several methods of IMRT. The dose distribution in the ipsilateral breast and the doses to OAR in plans for E-IMRT, T-IMRT, CP-IMRT and NCP-IMRT were compared with those for women who had previously received SWT. All IMRT plans showed some improvement in the parameters for breast distribution and/or dose to OAR when compared to SWT. These findings are consistent with previously published reports.
At present, although WBRT improves overall survival, the major problems are the long-term effects on the heart, lung and Contra Br. due to unwanted irradiation to these organs using SWT. In part, these can be minimized by (a) use of dynamic wedges (rather than static wedges) in the medial tangents to reduce the dose to the Contra Br. and (b) a half-beam block technique on the posterior borders of the tangential fields to prevent beam divergence into the lung and with CT planning to avoid dose to the heart for left-sided cancers.
Attempts have also been made to reduce breast tenderness thought to be due to fat necrosis in ‘hot spots’ within the dose distribution to the treated breast. Methods such as field-in-field dose modifications to even out the dose have been used, but this technique uses forward rather than inverse planning and by definition is not IMRT. Inverse planning is not a prerequisite for breast IMRT however, and forward planning may also be utilized.25
In our study we found small, but statistically significant, improvements in the parameters used to assess breast dose distribution. However, these are likely to have little clinical significance and a reduction in breast tenderness has not been reported using these changes in the planning technique.26
Hence we evaluated the IMRT plans on the basis of their effect on the OAR and the potential to reduce cardiac deaths and lung cancer as reported in the Overview study.27 Likewise, exposure of the breast to ionizing radiation in young women has been shown to increase the lifetime risk of developing a subsequent breast cancer.28,29 Hence, minimization of the dose to the Contra Br. is also highly desirable.
Overall, the IMRT plans reduced (a) heart volumes exposed to 95% of the prescribed dose to 15–30%; (b) the mean lung dose to 80%; and (c) the dose to the Contra Br. to 60–78% of that obtained with the SWT plans. It should be noted, however, that the volume of heart receiving a high dose is very small, even with SWT (1.7%). Analysis of other dose points demonstrated an improvement in heart volume exposed to intermediate doses of 30 Gy with the CP-IMRT technique, and a trend towards significance with the NCP-IMRT technique. Importantly, no significant ‘low dose bath’ effect at 5 Gy was observed for all techniques except CP-IMRT. It became apparent during the design of a class solution for CP-IMRT that excess dose to the organs at risk was likely to be observed, and this was borne out in the results. Given that we only specified a single heart dose constraint at a high level, further improvements could potentially be seen with further lower dose optimization points specified.
As shown in Table 2 the impact on OAR doses can be summarized as (a) E-IMRT has no effect; (b) CP-IMRT generally increases the OAR doses; and (c) both T-IMRT and NCP-IMRT decrease the OAR doses. Table 3 illustrates the simple percentage of patients who were noted have an improvement in dose distribution or OAR dose for any particular technique.
Table 3. Percentage of patients showing improvement with intensity modulated radiotherapy (IMRT) plans compared to standard wedged tangents (SWT)
Dose mean (%)
Dose to maximum (%)
Dose to minimum (%)
Organs at risk
Heart (left-sided resultsreported only)
Lung, mean dose (Gy)
Contralateral breast mean dose (Gy)
In our initial assessment of planning times we found NCP-IMRT more difficult and time consuming than T-IMRT. In addition, fluence checks are required prior to the first IMRT treatment and set-up accuracy is likely to be an issue with the NCP- IMRT as standard EPID is unlikely to be suitable as a verification method.
Our approach to whole breast IMRT requires a CTV to be defined in order to allow inverse optimization of beam fluences, whereas the E-IMRT technique does not require a CTV to be defined. It should be noted that several other groups have implemented a whole breast IMRT technique not requiring CTV delineation, although the planning techniques require specialized algorithms.30 A further use of IMRT that we have not investigated here is all-in-one treatment of the left breast and internal mammary nodes; one example of this technique utilized seven to 11 beams, and was, therefore, felt to be outside the scope of our study as is not a simple technique that can be applied to most patients.31
We have chosen to introduce T-IMRT for WBRT. This technique is conceptually the same as SWT, but does not have the problems cited above and importantly reduces doses to heart, lung and the Contra Br. After an initial learning period, the planning time for T-IMRT is unlikely to be increased over SWT, and for the same reason, patient set-up and linac time is likely to be equivalent to SWT.