British Journal of Radiology (2006) 79, 734-739
© 2006 British Institute of Radiology
doi: 10.1259/bjr/80814021
What is the optimum breast plan: a study based on the START trial plans
K Venables, PhD, MIPEM1,
E A Miles, MPhil, DCR(T)1,
E G A Aird, PhD, FIPEM2 and
P J Hoskin, FCRP, FRCR1
1 Marie Curie Research Wing, 2 Medical Physics Department, Mount Vernon Hospital, Rickmansworth Road, Northwood, Middlesex HA6 2RN, UK
 |
Abstract
|
|---|
Each year thousands of women within the UK are treated with radiotherapy for breast cancer. The majority of these women are treated using a medial and lateral tangential field. This study evaluates the plans submitted to the quality assurance (QA) team of the START trial and investigates some of the differences between departments. Throughout the START trial, hardcopies of the radiotherapy dose distribution on the central slice for one in three women were submitted to the QA team for analysis. The QA team measured physical parameters including breast size and lung depth as well as noting parameters used for the radiotherapy delivery including beam energy, field size and wedge angle. Over 1400 plans from 36 centres were analysed. The mean patient separation was 19.7 cm (SD 2.7 cm) with a mean lung depth of 1.5 cm (SD 0.7 cm). The modal beam energy was 6 MV and the mean wedge angle was 23°. Significant differences in the choice of wedge angle between departments were noted; however, in 90% of cases the resultant plan complied with the maximum dose gradient of 10% on the central axis specified by the trial protocol. Less than 3% (37 plans) had dose gradients of greater than 12%. This resulted in a mean dose gradient for all patients on the central axis of 5.7% (SD 2.9%).
|
Introduction
|
|---|
Many women with breast cancer are treated with radiotherapy each year. Guidelines on the production of isodose distributions have been given by the ICRU [1]. The START trial is a multicentre UK trial of breast radiotherapy comparing different fractionation regimens that accrued 4451 patients between January 1999 and December 2002 [2]. There were two randomization options called A and B. Trial A was a three way randomization to either 50 Gy in 25 fractions, treated daily over 5 weeks, or one of two schedules treating five times per fortnight to a total dose of 39 Gy or 41.6 Gy in 13 fractions of 3.0 Gy or 3.2 Gy, respectively. Trial B patients were randomized to 50 Gy in 25 fractions over 5 weeks or 40 Gy in 15 fractions over 3 weeks, treated daily. Guidelines on the production of isodose distributions were provided in the trial protocol, which stated that two tangential beams, angled to remove divergence in the lung, should be used. Departments were asked to use wedges to achieve a dose gradient of less than 10% across the breast on the central axis slice. In addition, cobalt-60 was only permitted for patients with a separation of less than 18 cm.
The trial has had a quality assurance (QA) programme from the outset. One of the aims of the QA team was to document the treatment technique in each centre and to ensure compliance with the protocol.
The treatment of breast cancer with radiotherapy is evolving and there is an increased awareness of the need for compensation in three dimensions. Within the START trial, only two centres used compensation other than a wedge. For some women, particularly those with larger breast sizes, this may be necessary to achieve dose uniformity particularly when doses away from the central axis are considered. Areas of increased dose within the breast may be the cause of breast pain and result in poorer cosmesis [3], while irradiation of large lung volumes could potentially lead to radiation pneumonitis [4]. In current clinical practice the majority of patients are treated with two opposing tangential fields. This paper analyses the data for these patients.
|
Method
|
|---|
Radiation dose distributions for 1 in 3 of the patients, assigned at the time of randomization, were collected by the QA team. Hardcopy central axis distributions were sent to the QA team after the patient had completed treatment. From these, data on the physical shape of the patient's breast, the linac parameters used for treatment and the resulting isodose distributions were collected. Data from each plan were checked by a second member of the QA team. Universal wedges were converted to an equivalent wedge angle using the formula:
where w1 is the weight of the wedge beam and w2 the weight of the corresponding open field. The wedge angle of the universal wedge was assumed to be 60° for all departments. During the analysis a further parameter of patient shape was defined:
Dose gradient was defined as the difference in dose between a minimum point dose and maximum isodose encompassing an area of 2 cm2. Unless there was a cold spot in the centre of the breast, departments were asked to record the minimum dose at a point 0.5 cm above the lung boundary on the perpendicular bisector of the posterior beam edge and at least 1 cm from the 50% field edge. This point was chosen so as to be in breast tissue rather than ribs, and to reduce the differences due solely to the use of different planning systems, which vary in their ability to account for lack of lateral scatter from the lung.
The QA team visited each department to observe simulation, planning and treatment of patients and to perform measurements on breast shaped phantoms. The results of the measurements in phantoms have been previously published [5, 6]. Questions relating to planning policy were asked of the staff producing the isodose distributions. Of relevance to the work presented in this paper were the criteria for their planning distribution, such as the choice of wedge angle, the resultant position and size (area and intensity) of hot spots and density correction used for lung. In a number of cases, this changed during the trial as better simulation facilities, particularly CT scanners, became available.
All analyses were performed using the SPSS program (SPSS Inc., Chicago, IL). Correlation (Spearman rank coefficient used if data not normally distributed) was used to test associations between dose gradient, wedge angle, breast shape and depth, patient separation, energy and wedge angle. Differences in dose gradient by centre were compared using analysis of variance (ANOVA). Data are presented using boxplots and scatter diagrams. In boxplots, the horizontal line shows the median value of the data, the length of the box represents the interquartile range, and the length of the line indicates the range excluding outliers and extremes (outliers/extremes are values more than 1.5/2.5 box-lengths from the 25th percentile).
|
Results
|
|---|
1488 plans were collected from 36 departments (1 centre input patients from 2 sites; these were counted as separate departments by the QA team since the planning systems were different on the two sites). Patient characteristics and treatment techniques are shown in 
Tables 1 and 2. The depths of lung incorporated into the treatment plan varied between departments and is shown in Figure 1.
A range of energies from Co60 to 10 MV was used for the treatment of these patients. The majority were treated using linear accelerators with a nominal energy of 6 MV. Energy used for treatment is illustrated in Figure 2. The energy used only correlated weakly with the patient separation (Spearman correlation coefficient 0.21 p<0.001) (Figure 3).
View larger version (22K):
[in this window]
[in a new window]
|
Figure 3. Association between energy used for treatment and average patient separation. Lines give the range of patient separations treated at each energy.
|
|
Of the plans received by the QA team, 748 incorporated a lung correction using either a standard value for all patients, which varied from 0.2 to 0.33 relative to water depending on centre, or bulk density corrections derived from CT data for an individual patient, which varied from 0.19 to 0.5 (53 plans). Ten patients were planned using a pixel by pixel correction for inhomogeneity.
The START trial stated that the recommended dose gradient across the patient breast be less than 10% on the central axis and 1390 plans (93.4%) complied with this. A further 95 plans (6.4%) exhibited dose gradients of between 10% and 15%. The remaining 3 plans (0.2%) had a dose gradient of greater than 15%. ICRU guidance suggests that plans should comply with a dose of 95–107% prescription dose over the target volume [1]. It was only possible to assess central sections and of these, 1232 plans (83%) complied with these recommendations. Dose gradient was found to be weakly correlated to breast depth (r = 0.50, p<0.001) (Figure 4) and with patient separation (r = 0.57, p<0.001) (Figure 5). This latter figure shows data separated by energy; as expected, the dose gradients are less if higher energy machines are used. The correlation between gradient and separation also improves if each energy is considered separately, and this is shown in Table 3.
View larger version (11K):
[in this window]
[in a new window]
|
Figure 5. Relationship between dose gradient and separation. Data for 6 MV have been excluded from the graph to allow the trends for other energies to be seen.
|
|
The mean wedge angle used was 22° for patients after lumpectomy (range 0–60°) and 33° (range 0–60°) for patients post-mastectomy. 690 patients were treated on linear accelerators with external fixed wedges (although a small number of these patients may have been treated using enhanced dynamic wedge (EDW). No significant difference was found between the medial and lateral wedge angles. A correlation was found between wedge angle and patient shape, which was stronger for some departments than for others. A department with strong correlation is shown in Figure 6 and the results for all patients in the trial in Figure 7. For centres where only fixed wedge angles were available, less correlation is observed. Wedge angle and patient shape did not correlate with an increased uniformity of dose gradient for the same centre. There was a negative correlation between wedge angle and dose gradient (medial wedge r = –0.35, p<0.001, lateral wedge r = –0.34, p<0.001). Differences in wedge angles used were also seen between departments.
View larger version (7K):
[in this window]
[in a new window]
|
Figure 6. Correlation between wedge angle and breast shape(max. distance from post edge to skin/patient separation) for centre 12. This was a large centre with a wide range of planning staff, but with rigid guidelines on the final distribution (r = –0.81 p<0.001, n = 44).
|
|
View larger version (20K):
[in this window]
[in a new window]
|
Figure 7. Correlation between wedge angle and breast shape for all patients in the trialr = –0.39, p<0.001. Note the wide spread in breast shapes for the fixed wedge angles 0°, 15°, 30°, 45° and 60°.
|
|
Significant differences were seen in the dose gradients obtained from different centres (ANOVA p<0.001). Further analysis of these differences revealed that even when comparing distributions from different departments with the same planning system and same nominal energy of accelerator (6 MV), differences were evident, as shown by Figure 8. Of particular note is the difference between centres B and E, both of whom put large numbers of patients into the trial. Centre B did not apply any lung correction but allowed for the effect of lung by planning with the apex of the breast hotter than the medial or lateral edges. Centre E estimated a lung shape based on the maximum lung depth and applied a lung correction of 0.25. The dose gradient recorded is not affected by the incorporation of lung correction. The planning system was a simple beam library system and the reduction in dose at the minimum point due to the loss of lateral scatter when the lung was included would not be reflected in the calculation. The lower than average dose gradient for centre B might suggest that they were not adequately accounting for the effect of the lung and were using steeper wedge angles than would have been used on a lung corrected plan. The higher than average dose gradients recorded by centre E might suggest that their estimate of lung shape was not sufficiently accurate and that they were overestimating the volume of lung in the plan.
View larger version (15K):
[in this window]
[in a new window]
|
Figure 8. Box plot showing difference in dose gradient for different centres using the same planning system and same nominal machine energy. Circles represent outliers.
|
|
The distribution of radiation within the breast was classified by comparing the dose at the apex of the breast (taken as 1.5 cm below the skin surface, Figure 9) with the dose in the most medial and lateral parts of the breast at least 1 cm from build up or penumbra regions. The majority of patients (70%) had an even distribution of radiation between these areas. However, in 11% of plans the medial and lateral aspects of the breast were hotter than the apex of the breast by 2–4%. When only those patients who do not have a lung correction are considered, the proportion in which the medial and lateral aspects were hotter than the breast apex was slightly less (8%).
|
Discussion
|
|---|
There was no correlation between average lung depth for a centre and average separation for that centre, which may have been observed if practice in some centres placed the lateral border more posteriorly than others. The lung depth included in the treatment field is affected by the patient's position; thus it would be necessary for some clinicians to include more of the lung in order to comply with the protocol recommendations on the positioning of the borders. In 1988, Fraass et al recommended the incorporation of lung density correction [7]. This is common practice in the UK with just over half of centres (51%) incorporating a correction and the majority of others planning to compensate for the effect of lung.
The weak correlation between energy and patient size may be related to the availability of different energies. In departments where high energy machines were available for any patient, there may be greater tendency for its use – one department treated 53% of patients using 10 MV photons. The depth of the 95% isodose increases as higher energy machines are used. There will therefore be a decrease in the coverage of the superficial regions of the planning target volume (PTV) when 10 MV photons are used compared with 6 MV photons. The depth below the skin surface at which the PTV should be drawn is usually taken to be 0.5 cm, however, compromise in this region is often preferred to decreased dose at depth. The exception is post-mastectomy patients for whom the loss of dose in the superficial tissues, when treated at high energy, represents a significant percentage of the planning target volume. Four patients in the trial who had undergone a mastectomy were treated using 10 MV photons without the application of bolus.
For the majority of machines in clinical use in the UK, the wedges are housed within the treatment head. However, some patients are treated on units with externally mounted wedges. The scatter from these external wedges can increase the dose to the patient's skin and contralateral breast and therefore it has been recommended by some authors that the wedging be applied predominantly from the lateral beam [8]. This practice was not seen in the START data where the majority of patients had the same wedge used for both medial and lateral fields. Differences were seen in the wedge angles used in different departments for the same "shape" of breast. The wedge angle needed will be affected by the energy of machine used. However, 75% of departments treated more than 80% of their breast patients at 6 MV and thus comparisons between departments are possible. When using fixed wedge angles there is less scope for optimization of the plan, whilst maintaining rapid treatment times. Departments using large wedge angles tend to produce plans in which there is more dose towards the medial and lateral aspect of the breast (see Figure 9
for definition of points). In addition, where lung correction has been omitted, true dose in these regions may be higher than indicated on the plan, due to the increased transmission through lung, leading to a possible increase in the risk of rib fracture.
Care must be taken in comparing distributions between planning systems due to the subtle differences in the implementation of the algorithms. However, differences in average dose gradient were also evident when comparing departments using the same planning system and same nominal energy of machine. No further statistical analysis of this aspect was performed because of the small numbers in some of the groups, but it reflects differences in planning policy. Some departments with conventional rather than CT simulators estimated the lung shape using either a rule of thumb or a lung template. Although this will usually give a more realistic estimate of the distribution than ignoring the lung correction completely, it may result in a different distribution from that which would be produced if more accurate information was available.
In some instances plans are produced in which no attempt is made to estimate the lung outline. In these cases it is not possible to incorporate lung correction to compensate for the reduced attenuation of the lung relative to breast tissue. One way to overcome this is to produce a plan which indicates an artificial dose gradient across the breast that will not be present in the patient. For these patients, it is expected that plans sent to the QA team will aim for the dose at the apex of the breast on the plan to be hotter than that at the medial and lateral border. For a small percentage of patients, the reverse of this situation was seen and the medial and lateral aspect of the plan was hotter than the apex. In a few cases, this may be necessary to compensate adequately for variations in the shape of the breast in three dimensions, but should not be normal practice if only the central axis of the breast is considered.
For the majority of patients, only data on the central slice was sent to the QA team; this will mask the true dose gradients which will be present in the patient's breast as a whole. For some women to conform with ICRU recommendations of a maximum dose gradient of 95–107%, intensity-modulated radiotherapy will be necessary to eliminate hot spots in the superior and/or inferior aspect of the breast.
|
Conclusion
|
|---|
The analysis of the plans submitted to the QA team emphasises the high degree of protocol compliance that was achieved for this trial. The collection of central axis distributions by the QA team enabled protocol violations to be flagged up to centres early in the trial and thus improved the compliance with the protocol. It also enabled the QA team to discuss alternative patient positioning where larger than average lung depths were seen. The data collected represents a snapshot of clinical practice in the years the trial was accruing and will aid the analysis of the main trial once mature survival, recurrence and late effect data are available.
When producing a dose distribution for a patient, a number of factors must be considered. Ideal breast plans have a 95% isodose which covers breast tissue to within 5 mm of the skin surface. No hot spots in excess of 107% should be present (lower if only a single slice is assessed) and the hotspots should be located evenly throughout the breast tissue. Ideally, patients should have CT scans to obtain accurate information on the position of the lung. If this is not possible, the effect of low density lung tissue should be carefully considered and the distribution should appear hotter at the apex of the breast. For some patients, compromise must be reached and the planner should carefully weigh the increased penetration available by using higher energy beams against the decrease in dose to superficial tumours. In particular, if post-mastectomy radiotherapy is to be given using photons with energies higher than 6 MV then the use of bolus should be considered.
|
Acknowledgments
|
|---|
The START trial management group: Edwin Aird, Jane Barrett, Peter Barrett-Lee, Judith Bliss, Jackie Brown, John Dewar, Jane Dobbs, Jo Haviland, Penny Hopwood, Peter Hoskin, Pat Lawton, Brian Magee, David Morgan, Roger Owen, Eileen Parkin (RAGE Observer), Joyce Pritchard (RAGE Observer), Val Speechely, David Spooner, Mark Sydenham, Karen Venables, Elizabeth Miles and John Yarnold.
Received for publication October 24, 2005.
Revision received December 16, 2005.
Accepted for publication January 9, 2006.
|
References
|
|---|
- International Commission on Radiation Units and Measurements. Prescribing, recording and reporting photon beam therapy. ICRU Report 50. Bethesda, MD: ICRU, 1993
- START trial management group. Standardization of Breast Radiotherapy (START) Trial. Clin Oncol 1999;11:145–7.
- Yarnold JR, Donovan EM, Bleakley NJ, Reise SF, Regan J, Denholm E, et al. Randomised trial of standard 2D radiotherapy (RT) versus 3D intensity modulated radiotherapy (IMRT) in patients prescribed breast radiotherapy. Radiother Oncol 2002;64:S15
- Early Breast Cancer Trialists' Collaborative Group. Favourable and unfavourable effects on long-term survival of radiotherapy for early breast cancer: an overview of the randomised trials. Lancet 2000;355:1757–70.[CrossRef][Medline]
- Venables K, Winfield EA, Deighton A, Aird EGA, Hoskin PJ. The START Trial – measurements in semi-anatomical breast and chest wall phantoms. Phys Med Biol 2001;46:1937–48.[CrossRef][Medline]
- Venables K, Winfield EA, Aird EGA, Hoskin PJ. Three-dimensional distribution of radiation within the breast: an intercomparison of departments participating in the START trial of breast radiotherapy fractionation. Int J Radiat Oncol Biol Phys 2003;55:271–9.[CrossRef][Medline]
- Fraass BA, Lichter AS, McShan DL, Yanke BR, Diaz RF, Yeakel KS, et al. The influence of lung density corrections on treatment planning for primary breast cancer. Int J Radiat Oncol Biol Phys 1988;14:179–90.[Medline]
- Ikner CL, Russo R, Podgorsak MB, Proulx GM, Lee RJ. Comparison of the homogeneity of breast dose distributions with and without the medial wedge. Med Dosim 1998;23:89–94.[CrossRef][Medline]
This article has been cited by other articles:
|
|
|
|
 
E J ADAMS and A P WARRINGTON
A comparison between cobalt and linear accelerator-based treatment plans for conformal and intensity-modulated radiotherapy
Br. J. Radiol.,
April 1, 2008;
81(964):
304 - 310.
[Abstract]
[Full Text]
[PDF]
|
|
|
Top
Abstract
Introduction
