br This study describes the estimation of
This study describes the estimation of cardiac segment doses for women who received 2-dimensional planned breast cancer regimens and subsequently developed segment damage. We describe the interregimen variation in segment doses for 41 regimens and interpatient varia-tion in segment doses for 14 commonly used regimens.
We considered how these segment doses may be used to assess associations between dose and segment injury. Our results were used to inform a separate study that related segment doses to sites of injury.17
Methods and Materials
Regimens were identified from the radiation therapy charts of 470 women included in a population-based study of major coronary events with known location of segment injury after breast cancer radiation therapy.3,17 The women were irradiated in Sweden between 1958 and 2001 or Denmark between 1978 and 2000. Radiation therapy charts included diagrams or photographs of the treatment fields, and sometimes dose-plans. Details on the surgery, target definition, field borders, target dose, applied total dose, dose per fraction, beam energy and use of shielding, wedges, and Corn oil were collected. In-formation was also collated from radiation therapy protocols.
Ten radiation therapy CT-planning scans were randomly selected from women irradiated at Odense University Hospital in Denmark in 2010. The treatment position was supine, with both arms above the head. The scan slice thickness was 3 mm, and intravenous contrast was not used. The whole heart, ventricles, and LV and coronary artery segments were contoured on all 10 scans using an atlas.18 To simulate a mastectomy, the breast was contoured and assigned a CT-value for air and 1 cm of tissue was retained above the pectoralis major muscle to account for residual subcutaneous tissue.
Selection of typical computed tomography scan
The 2 most common left-sided regimens for the women in the study were identified and reconstructed on all 10 CT scans, which were a midline tangential regimen used after breast conserving surgery (Fig. 1A; Table E1 [field arrangement 3]; available online at https://doi. org/10.1016/j.prro.2019.01.004) and a direct electron chest wall regimen used after mastectomy (Fig. 1E; Table E1 [field arrangement 7]). Whole heart and cardiac sub-structure doses were collated, as were whole heart vol-ume, chest wall separation, sternal length, and Haller index (ie, ratio of height between anterior spine and posterior sternum to transverse width of the chest). The scan with the mean heart doses closest to average for both techniques, and which was not atypical for any of the anatomical factors examined, was selected as the typical
Reconstruction and dose calculation
All identified regimens were reconstructed on this typical CT scan using 3-dimensional treatment planning (Varian Eclipse Treatment Planning System, version 10.0.39). Lines were drawn on the body surface to represent clinical landmarks used to plan 2-dimensional radiation therapy in previous decades. Field borders, gantry angles, and custom blocks19 were guided by these clinical surface markings and digitally reconstructed radiographs.
The dose calculation algorithms were an analytical anisotropic algorithm for photon plans, Monte Carlo for electron plans, and pencil beam for cobalt plans. If photon beam energies were unavailable, they were created using mixed energy beams. The dose was calculated using the 0.1 cm calculation volume grid for all, except the cobalt regimens where the minimum grid was 0.25 cm. Dose-volume histograms (DVHs) were exported for each car-diac segment with dose-bins of 0.1%.
Orthovoltage regimens were reconstructed by manual planning. Field borders were defined using CT-based 3-dimensional virtual simulation. Ten axial CT images (CT slice spacing: 1.2 cm) spanning the heart were printed and scaled up to life size. Isodose charts were superimposed onto each CT slice and used to map dose onto cardiac segments. The isodose shift method was used to correct for lung in the field(s) and standoff at the body surface (isodose shift factor: 0.8). The proportions of each cardiac segment included within the isodose lines were calcu-lated. DVHs were plotted and mean segment doses calculated.
For all regimens, DVHs were used to calculate the mean doses in equivalent 2 Gy fractions, which was separately calculated for each dose bin in each DVH using:
where n Z number of fractions, d Z mean dose to the cardiac structure per fraction (Gy), and a/b Z 2 Gy.20,21
Interpatient dose variation
The effect of interpatient variability in anatomy on cardiac segment doses was investigated in 14 regimens (with different field arrangements) from the 2 most common technique categories. Eight of these 14 regimens (4 left-sided and 4 right-sided) were tangential (Table E1; Table 1 [techniques 1-4]) and 6 (3 left- and 3 right-sided) were anterior electron regimens (Table E1; Table 1 [techniques 5-7]). The left regimens were reconstructed on all 10 CT scans, and the right regimens on only 5 scans