Computed tomography for excision cavity localization and 3D-treatment planning in partial breast irradiation with high-dose-rate interstitial brachytherapy

Materials and methods 

PTV and implant geometry definition 

Before implantation 

For target volume localization the breast is scanned by CT (5mm slice thickness and step in the first 17 patients, 2.5mm in the other 37 because 2.5mm slice thickness and step more accurately define upper and lower cavity edges). The patient is placed in the supine position, with the ipsilateral upper arm extended to the trunk and the forearm bent so the palm of the hand rests upon the head. The same arm position must be used during implantation, post-implant CT and therapy. To ensure position reproducibility 2 points are tattooed at the mediosternal and medium axillary lines. Images are transferred to a treatment planning system (TPS) for external beam radiotherapy (RT) (Pinnacle3 – Philips). The excision cavity is outlined on each slice and, using the TPS volume expansion feature, expanded by 1–2cm to provide the planning target volume (PTV), which is then modified to ensure at least 5mm distance from skin and chest wall.

Expansion varies with breast dimension, lumpectomy cavity location, size and shape. The smaller the breast, the less expansion there must be to avoid overdose to the skin and/or the chest wall and healthy breast tissue. To prevent damage when the cavity is large or close to the chest wall or the skin, expansion might need to be under 1.5–2cm. Depending on the location of irregularly shaped cavities in relation to the skin or chest wall, expansion will probably have to vary at different points. Furthermore, cavity expansion also depends on histological findings as the tumour is not always in the centre of the surgical explant [4]. Consequently, the larger the tumour-free resection margin, the smaller the PTV safety margin. For example, if histological findings report a 3mm margin on one side and a 10mm margin on the other, expansion should be greater on the 3mm side.

Once the PTV has been obtained, a virtual simulation of the treatment plan is carried out using external beam RT software to provide two medial and lateral opposed half beams, the dorsal planes of which become the implant dorsal plane. Medial and lateral digitally reconstructed radiographs (DRRs) of the beams, displaying the PTV, are printed on a scale of 1:1 (Figs. 1A and B). A template is placed over the two DRRs to determine implant needle entrance and exit points.

View Large Image Download to PowerPoint

Fig. 1. Digitally reconstructed radiographs in which planning target volume (PTV) is displayed. Needle entrance and exit points are determined by placing a template over the PTV. (A) Medial: red=reference point; blue=beam central point.(B) Lateral: green=reference point; orange=beam central point.

Implantation day 

The patient is taken to the simulator. Using the two tattoos as landmarks, the central points of the dummy beams are drawn on the patient’s skin. The dummy beam dorsal planes are then projected and drawn on the skin; needle entrance and exit points are marked (Fig. 2) by placing the template on the patient’s skin.

View Large Image Download to PowerPoint

Fig. 2. Dummy beam dorsal planes are projected and drawn on the patient’s skin. Entrance and exit points of the superior and inferior needles, previously defined by superimposing a template on the digitally reconstructed radiographs, are marked by placing the template on the patient’s skin.

During implantation, superior and inferior needles on the dorsal plane are inserted by hand, using the markers on the skin as guides. The other needles are inserted using a triangular plastic template (16mm catheter separation; approximately 14mm interplane separation) and a metal template holder to provide accurate needle placement and precise implant geometry. After needle placement, template and holder are removed and the needles are replaced with flexible catheters equipped with buttons blocking both ends.

Results 

This approach for localizing the PTV and designing implant geometry was evaluated in 54 patients. A two-plane implant was constructed in 11 cases and a three-plane in 43; median number of implanted catheters was 12 (range 7–15). In all patients post-implantation CT images visualized precise catheter placement with respect to the lumpectomy cavity, so no adjustments were required and the virtual implant geometry was confirmed as accurate.

In the first 8 patients, catheter reconstruction was based on two orthogonal films because the CT-based 3D-treatment planning software was not yet available at our Institution. It was used from September 2004 in the last 46 patients and provided more precise and easier catheter reconstruction, a shorter catheter reconstruction time and more accurate measurements of active and inactive source lengths. Furthermore, the CT-based 3D software visualized dose distribution in the PTV and in organs at risk and calculated the dose–volume histograms. A two-plane implant was constructed in 10 cases and a three-plane implant in 36; the median number of implanted catheters was 12 (range 7–15). Median V100value was 106.0cc (range 43.6–187; mean 106.2; standard deviation – SD – 35.1). Median DHI was 0.75 (range 0.68–0.81; mean 0.76; SD 0.03). Frequency distribution of DHI is reported in Table 1. The mean dose covering 90% of the PTV was 3.86±0.41Gy (96.4%±10.17% of prescribed dose; range 3.08–4.40; median 4.07Gy). Median CI was 1.97 (range 1.50–2.62, mean 1.99, SD 0.46) (Table 2). A mean of 12.39% of healthy breast tissue received 100% of the prescribed dose (range 8.95–15.8, SD 2.58, median 12.75%). Dose distribution around the PTV in a patient with a three-plane implant is shown in Fig. 3.

Table 1. Dose homogeneity index (DHI): frequency distribution in 46 patients in whom a CT-based-3D planning system was used

	DHI	Number of patients
	0.68–0.70	6
	0.71–0.73	5
	0.74–0.76	19
	0.77–0.79	14
	0.80–0.82	2

Table 2. Dosimetric results in 46 patients in whom a CT-based-3D planning system was used

	Parameter	Median	Mean	Range	SD
	PTV volume (cc)	57.9	57.6	20.6–106.0	21.7
	V50 (cc)	276.5	280.0	142.0–455.5	81.6
	V100 (cc)	106.0	106.2	43.6–187	35.1
	V150 (cc)	27.2	25.7	10.6–40.7	6.5
	V200 (cc)	12.0	11.9	5.3–27.2	3.7
	DHI	0.75	0.76	0.68–0.81	0.03
	D90 (Gy)	4.07	3.86	3.08–4.40	0.41
	CI	1.97	1.99	1.50–2.62	0.46

SD, standard deviation; PTV, planning target volume; DHI, dose homogeneity index; CI, conformity index.

View Large Image Download to PowerPoint

Fig. 3. Relative dose distribution of a three-plane implant with PTV on an axial CT slice.

Discussion 

When high-dose-rate brachytherapy is used for PBI precise pre-implant localization of the target volume and definition of implant geometry are required [2], [14], [21]. The pre-implant CT-based procedure we developed is accurate because post-implant CT images visualized correct catheter positions with respect to the PTV in all patients so the implant required no adjustments.

CT scanning provides the best level of 3D accuracy in visualizing the surgical cavity [2], [13], [21] for up to 6–8 weeks after surgery [21], i.e. the time-span within which PBI should be delivered. Different CT-image based techniques have been developed for designing a virtual implant geometry before implantation. Perera et al. [15] and Vicini et al. [20] used radiopaque angiocatheters as landmarks for needle entrance and exit points on the breast in a virtual brachytherapy technique. They achieved excellent agreement between pre- and post-implant geometry and no implant adjustment was required in any patient. A more complex, extremely precise technique, described by Cuttino et al. [5], avails of a single section in the CT-simulator suite. After a breast CT scan, the target volume is delineated and a 3D-treatment planning is generated. Catheters are positioned freehand under CT guidance. The entire procedure takes less than 2.5h but cannot be widely adopted because anaesthesia support and a CT simulator are not always available.

Our technique defines implant geometry and catheter placement precisely with respect to the PTV using only two tattoos as landmarks at the mediosternal and medium axillary lines rather than a set of angiocatheters which have to be left in place until the virtual geometry implant has been designed [15], [20]. The tattoos do not cause any discomfort and also serve to reproduce the patient set-up during implantation and subsequent treatment sessions, thus preventing changes in breast position and catheter misalignment. The virtual implant is constructed with external beam TPS and field gantry and collimator angles adjust the implant geometry to the patient’s anatomy. The PTV is localized in the digitally reconstructed radiographs of virtual medial and lateral fields. Guided by the tattoes, the dummy beam central points, the implant dorsal plane and needle entrance and exit points are marked on the patient’s skin.

Like other techniques [5], [15], [20] our simple user-friendly method provides similar results in terms of correct catheter placement with respect to the PTV. As breast interstitial brachytherapy is an operator-dependent technique requiring long training, it is not widely available. CT-based virtual pre-implant geometry construction, such as we describe, renders the procedure less operator dependent and more user-friendly even for radiation oncologists without extensive training in breast brachytherapy.

Treatment quality is guaranteed by precise post-implantation PTV localization, which leads to precise catheter reconstruction, exact inactive and active source length measurement and optimization of dwell weights and positions. All this can be achieved only with image-based 3D information [2], [5], [6], [12], [14].

Treatment planning with CT-based 3D software can provide homogeneous dose distribution in and around the PTV. In our series median D90 values of 4.07Gy show we had achieved good PTV coverage in line with other reports. In Das et al. a median of 96% of target volume received 100% of prescribed dose [6], while Cuttino et al. reported that a mean of 95% PTV (defined as the lumpectomy cavity plus 2cm) received >90% of prescribed dose [5].

A template-guided implant, even with three planes, may not achieve good PTV coverage due to template size [20]. The excellent results obtained by Das et al. who used a template but inserted needles freehand in regions not covered by the template are linked to implantation being performed under cavity visualization [6]. In our series PTV coverage was good. In fact we did not use high-dose-rate brachytherapy for PBI in patients with a large PTV because when we placed the template over the PTV on DRRs we found it did not provide good geometric coverage. Consequently, our mean V100 is smaller than those reported by others (106.2cc vs 211cc achieved by Cuttino et al. and vs 226cc by Das et al.) [5], [6]. We must admit the PTV/V100 ratio is higher than it should have been. Since we used only one type of template to cover PTV we had to irradiate a larger volume of healthy tissue. However, even though the CI was almost double the optimal, it still fell within the range in which treatment is considered to comply with the treatment plan [8]. The disadvantages of the single template will be overcome by using a set with different interplane separation.

DHI is an index of local hot spots in the implant which have been related to fibrosis and fat necrosis [23]. To ensure individual implant quality and homogeneous dose distribution DHI has to be ⩾0.75 and the median value was, in fact, 0.75 in our series (Table 1 for distribution).

Using CT-based 3D software one can visualize the dose distribution in the PTV and the organs at risk on CT slices and calculate the dose–volume histograms. Caution was exercised to prevent unnecessary irradiation of healthy tisssue. Dwell times were modified if too high dose was delivered to the skin so as to prevent teleangiectasis. Dose to the healthy breast tissue was in line with other report (12.39% vs 10% achieved by Weed et al.) [24]. Unlike external beam RT, administered heart and lung doses are far below the limiting value, as demonstrated in the literature [24].

In conclusion, in our series we achieved good coverage of the target volume and homogeneous dose distribution, as demonstrated by the D90 and DHI values. This is a consequence of correct catheter positioning in relation to PTV. In fact, if we had not exactly positioned the catheters, we would probably have been forced to modify dwell times at some dwell position to cover the target volume at the expense of dose homogeneity. We would have had more regions in the target volume receiving over 150% of the planned dose, as reported by Kestin et al. and Major et al. [10], [12]. This would have increased the probability of side effects which, although not life-threatening, impact upon cosmetic results and patient’s quality of life.

Finally, one type of template (with 14mm interplane separation in the present instance) clearly emerges as sub-optimal for all implants. To ensure optimal dose distribution on the PTV and to avoid irradiating healthy tissue unnecessarily, a set of different templates appears needed. Indeed in an ongoing investigation, we have adopted a case-by-case approach and now decide the best strategy for each patient: template with selected interplane separation, freehand, or a combination of template and freehand.