Journal of Medical Radiation Sciences

Patient selection

Twenty‐seven women treated during 2015 using IP‐IMRT WBRT were randomly selected from across two clinics: Radiation Oncology Centres, Gosford (n = 21), and Radiation Oncology Centres, Wahroonga (n = 6), and replanned using TD. All patients were simulated using customised vacbag immobilisation (Bionix, Toledo, OH) with both arms positioned overhead. Static free breathing scans were acquired for each patient: Gosford images with a GE kV CT scanner (GE Healthcare, Little Chalfont, United Kingdom) using 2.5‐mm slice intervals and Wahroonga images with a Somotom Definition Flash (Siemens Healthcare, Erlangen, Germany) using a 2‐mm slice thickness. The sample consisted of 6 left‐sided and 21 right‐sided WBRT cases.

Regions of interest

For each patient, a planning target volume (PTV) defining the whole visible breast was delineated by a radiation oncologist. The PTV was derived using cues from external radiopaque markers placed by radiation therapists at simulation as well as discernible breast tissue evident on CT images. The PTV was limited by the external contour minus 3 mm and the ipsilateral lung and heart plus 4 mm. Efforts were made to ensure PTV consistency practice wide.

The retracted PTV enabled it to function as a planning optimisation structure. When optimising to superficial volumes, optimisation functions will try to compensate for the lack of build‐up and lateral scatter on the surface by using high superficial photon fluences, particularly when exclusively tangential beams are used.21 This may result in unnecessarily high superficial monitor unit (MU) delivery, causing high skin doses.21 Retracting the PTV from the surface avoids this potential compromise in plan quality, as shown in Figure 1.

To account for intrafraction motion as well as random and systematic errors, the field aperture was widened to overshoot the PTV anteriorly by at least 2 cm. Radiation therapists added relevant OARs including the ipsilateral lung and heart. The heart OAR was confined to the pericardial cavity and excluded superiorly located major vessels, as per Radiation Therapy Oncology Group (RTOG) guidelines.23 A CTV was subsequently added around the lumpectomy site by a radiation oncologist.

IMRT technique

The IMRT planning technique consisted of two tangential open parallel opposed beams and two tangential IP‐IMRT parallel opposed beams utilising the same gantry angles. The plans were prepared in RayStation v4.5 (Raysearch, Stockholm, Sweden) for a conventional linear accelerator capable of delivering both 6‐ and 10‐MV photon beams. The isocentre was placed on the posterior edge of the beams to minimise integral penumbra and conformal beam angles chosen to reduce dose to normal tissue. The open beams delivered a minimum of 60% of the prescription dose to the PTV, with up to 95% delivered anteriorly due to the breast contour. The open beams maintained approximately 2 cm of flash to ensure adequate dose coverage in the event of target deformation and intrafraction motion. The energy chosen was either 6 or 10 MV, guided in part by patient separation (6 MV n = 12; 10 MV n = 15). The IP‐IMRT beams acted as an optimised wedge to preferentially increase the dose coverage of the PTV edge bordering the ipsilateral lung. IP‐IMRT beams consisted of 6‐MV static ‘step and shoot’ multi‐leaf collimators (MLCs) (Gosford, n = 21) or 6‐MV dynamic MLCs (Wahroonga, n = 6); 50 Gy in 25 fractions was prescribed with the essential goal of PTV D₉₉ > 47.5 Gy. Optimisation criteria were added with the aim of maximising target homogeneity and conformity while minimising OAR dose. The typical setup and dosimetry are depicted in Figure 2.

TD technique

Plans previously optimised with the IMRT technique (n = 27) were replanned using TomoTherapy's topotherapy platform: TomoDirect™ (TD) (Fig. 3). Identical medial and lateral beam incident angles were utilised to maintain setup consistency, with the only available (6 MV) beam energy utilised. TD parameters included a dynamic field width (FW) of 2.5 cm, a (default) pitch of 0.251 and a modulation factor (MF) of 2.000.

The FW is the collimator defined field size in the longitudinal direction; it can be fixed to one of three settings: 1, 2.5 or 5 cm (defined at the isocentre). Varying the FW will affect the sharpness of dose fall‐off in the longitudinal direction, as well as the modulation possible over the target. A smaller FW will increase treatment delivery times; a 2.5‐cm FW was chosen to achieve a balance between satisfactory treatment time and suitable modulation. The selection of dynamic jaws, or TomoEdge™, allowed the FW to reduce to 1 cm directly before and after the target. This recent hardware upgrade allowed for a reduction in longitudinal penumbra without a significant compromise to delivery time. The MF is an indicator of the degree of intensity modulation permitted, typically ranging from 1.4 to 3.0. A high MF can affect both treatment and planning time, since a higher degree of modulation will require more iterations to optimise. The chosen modulation factor of 2.000 reflected typical practice wide modulation. In TD mode, the pitch is the direct outcome of the FW; it is defined as the distance of couch travel in centimetres per sinogram projection.11

No OARs were included in optimisation, besides a dose control volume (DCR) of +0.2 cm to +2.0 cm placed along the posterior edge of the PTV. At least 2 cm of flash was included to ensure adequate dose coverage in the case of target deformation. Because TD fields are characterised by heterogeneous fluence, adding flash involved projecting the average fluence of the two closest leaves intersecting the PTV. This offered a significant dosimetric advantage over IMRT:24 the IMRT hybrid technique only provided flash from the open beam portion; TD flash provided optimised flash for the entire beamset.

Typical target optimisation parameters (Fig. 4) included the goal of 80% of the PTV receiving 50 Gy to ensure that the PTV D₉₉ exceeded 47.5 Gy. This starting point enabled quick iterative progress towards achieving the ICRU50 heterogeneity limits of +7% (53.5 Gy) and −5% (47.5 Gy).18 The DCR allowed integral dose limitation and was routinely set to 60% of the prescribed dose (30 Gy), with D₂ set to 28 Gy.

QA

TD plans underwent individual quality assurance (QA) using ArcCHECK (Sun Nuclear, Melbourne, FL), with a gamma tolerance of 3 mm and 3% and a threshold of 10%. Plans with less than 95% of points passing the gamma analysis were rejected and replanned to ensure the accuracy of treatment delivery. IMRT plans had their open beams verified using RadCalc (LifeLine Software, Austin, TX), with up to 5% calculated deviation deemed acceptable. IP‐IMRT beams were verified using ArcCHECK (Sun Nuclear, FL), with a gamma tolerance of 3 mm and 3% and a threshold of 10%, using the same pass criterion as TD plans.

PTV metrics

There was no significant difference found in the PTV D₉₉ > 47.5 Gy point of comparison (TD D₉₉ = 48.0 Gy vs. IMRT = 48.1 Gy, P = 0.26). Prescription dose (50 Gy) coverage was also statistically similar (TD V₅₀ = 85.5% vs. IMRT = 82.0%, P = 0.09) (Table 1).

PTV maximum doses were defined as the dose received by 1% of the PTV (D₁). This reporting approach rendered higher near maximum doses than the D₂ criterion of ICRU83,20 but was consistent with shared departmental protocols. The mean D₁ for the TD arm was 53.5 Gy compared to 53.0 Gy for the IMRT subset (P = 0.02). Variation was higher within the TD group (σ: 1.11 vs. 0.35), a phenomenon most likely due to the effect of patient separation and the unavailability of higher TD beam energies. TD plans were also found to have generated higher median doses (TD = 51.1 Gy, σ = 0.45 vs. IMRT = 50.9 Gy, σ = 0.38, P = 0.03). Nevertheless, the mean of both groups met ICRU50 target heterogeneity goals (+7% to −5%).18

IMRT plans demonstrated a statistically significant advantage in target homogeneity (HI TD = 0.11 vs. IMRT = 0.10, P = 0.03). While there are various HI formulae applied within literature, ICRU83 advocates the definition (D₂–D₉₈)/D₅₀.20 We modified this formula slightly, defining HI as (D₁–D₉₉)/50 Gy, primarily to maintain internal consistency given the utilisation of D₉₉ and D₁ data. Figure 5 illustrates the statistically significant differences of target metrics.

OAR metrics

TD plans produced a statistically significant reduction in V₅ ipsilateral lung doses (TD V₅ = 23.2% vs. IMRT = 27.2%, P = 0.04), of which clinical significance is uncertain. Other queried OAR metrics remained comparable (TD ipsilateral lung V₂₀ = 13.2% vs. IMRT = 14.6%, P = 0.30; TD heart V₅ = 2.7% vs. IMRT = 2.8%, P = 0.47; TD heart V₁₀ = 1.7% vs. IMRT = 1.8%, P = 0.44) (Table 2).

Optimisation time

The time required to reach sufficient dose optimisation was recorded in both study arms (Fig. 6). TD was found to reduce planning time considerably (TD = 9.8 m vs. IMRT = 27.6 m, P < 0.01) and reduce variation in planning time (TD σ = 2.77 vs. IMRT = 4.33). The considerable mean time saved per case (17.8 min) was compounded by the ease of data transfer from the TomoTherapy treatment planning system to the patient record and verify system.

The energy effect

Tomotherapy is limited to a single 6‐MV treatment energy. The lack of a higher energy option restricted the ability to lessen maximum (D₁) doses, particularly in patients with large PTV posterior edge (PE) separations. This limitation was reported by Fields et al., among others, who found that PE separations >21 cm often produced hotspots >110%.4, 25-28 Similarly Das et al.26 found a correlation between hotspots >115% and chest wall separations >22 cm.

A linear regression analysis of our own data revealed the extent to which higher energies (10 MV) had the ability to reduce maximum (D₁) PTV doses. Once PE separation reached 22 cm, the differential advantage of the 10‐MV option available in IMRT plans became evident, as demonstrated in Figure 7.

This data may be useful to screen patients who are potentially better suited to higher energy IMRT at the time of simulation.

Static beam angle versus helical techniques

The TomoTherapy HD platform allows for helical delivery over a rotational interval of 7°, up to a total of 51 projections for each gantry revolution. With excellent high‐dose conformity, Haciislamoglu et al.3 among others argue the merits of helical delivery in WBRT. Franco et al.11 state the obvious drawback of helical techniques: larger integral dose. This low‐dose bath has been shown to be the genesis of secondary malignancy, particularly in the contralateral breast.9 Static beam angles facilitate avoidance of tissues outside of the treatment path, with healthy tissues receiving only minor scatter radiation. This was demonstrated by our results, particularly in the low V₅ ipsilateral lung doses (TD V₅ = 23.2%; IMRT = 27.2%). We suspect this was partly due to the use of dynamic jaws, or TomoEdge™, which constrains longitudinal penumbra. Static beam angles are the safest approach when seeking to reduce risk in younger breast cancer patients, who potentially benefit only minimally from WBRT.11

A recognised benefit of static beam techniques is the reduction in treatment delivery time.11 Helical tomotherapy may employ the use of complete and directional ‘blocks’ to preclude radiation from specific volumes. These blocks result in prolonged treatment times due to the constraint of irradiation from certain treatment angles while the gantry rotational velocity remains constant. Anecdotally TD and IMRT treatment delivery times are similar. Factors such as required collimator rotation, modulation and use of either static ‘step and shoot’ or dynamic IP‐IMRT options may impact treatment delivery time. The choice of a 5‐cm FW would likely offer a small time advantage to TD, with a modest compromise to modulation capacity.

TD imaging involves a MVCT helical approach, and we found that the time taken to acquire a typical 3‐mm reconstructed image was approximately 160 sec. This exceeded the conventional linear accelerator (kVCT) imaging time of approximately 30 sec (left sided 280–90°; right sided 182–345°), placing TD at a disadvantage in terms of its ability to image for deep inspiration breath hold (DIBH) procedures (patients are typically unable to maintain a breath hold for much longer than 30 sec).29