Breast size and dose to cardiac substructures in adjuvant three-dimensional conformal radiotherapy compared to tangential intensity modulated radiotherapy

Cardiovascular diseases are becoming the most critical competing mortality risk in women with early breast cancer treated with present-day radiotherapy (RT).1,2, The relative risk of radiation-induced heart failure increases with rising cardiac radiation exposure, typically reported as mean absorbed dose to the whole heart (MWHD).3,4,5 MWHD values reflect local radiation therapy practices, and with the help of modern RT approaches, now ranging from 1.7–5.4 Gy6,7,8 and 1.22–1.65 Gy9, for mean and median values, respectively. However, even very low cardiac exposure does not eliminate the risk of radiotherapy-mediated cardiotoxicity, which has been demonstrated in recent studies.3,5,10

In many recent publications, authors favor the use of intensity modulated techniques over three-dimensional conformal radiotherapy (3D-CRT) in node negative breast cancer adjuvant RT, arguing for lower MWHD, decreased skin toxicity and more homogeneous dose distribution in the target volume.11,12,13 Besides the RT technique used, MWHD depends on the position of the patient’s heart relative to the irradiated breast and the shape of their chest wall.14 Different simple anatomical measures were evaluated to predict increased MWHD and subsequently for the need to use one of the heart-sparing techniques, namely deep inspiration breath hold technique (DIBH). Useful anatomical measures are increased chest wall separation (CWS)9, maximum heart distance (the distance between the anterior cardiac contour crossing over the posterior edge of the tangential fields)15, multidimensional assessment of the presence of the heart in contact with the chest wall14 and linear heart contact distance from the left sternal to the beginning of the lung parenchyma edges at the 4^th costal arch in the axial axis.16 It has also been shown that the shape and size of the clinical target volume (CTV) result in increased mean and/or maximum point heart doses.9,17,18 If a cohort of breast cancer patients with similar breast volume is defined, specific problems and resolutions can be proposed, because breast contours according to breast size and shape may be associated with the variations in the target volume coverage and calculated dose to organs at risk.19 Three-dimensional treatment planning allows target volume to be measured and CTVs of ≤ 500–975 cm³, 975–1.600 cm³ and ≥ 1.600 cm³ have been typically, but not consistently, defined as small, medium and large breasts, respectively.20,21 Additionally, quite a few clinical studies have reported a comparison of the clinical adverse events in regard to the three groups of breast sizes.22

Although observed average MWHD in a population of breast cancer survivors is low, smaller fragments of the heart might have received doses exceeding 25–40 Gy.4,10,23,24 Subclinical cardiac dysfunction was observed early after adjuvant radiotherapy for breast cancer with molecular bio-markers25,26 radionuclide myocardial perfusion imaging27,28,29, echocardiography30,31,32, and functional magnetic resonance imaging.31 Limited data exist regarding the range of doses received by individual heart substructures with adjuvant free-breathing 3D-CRT or tangential intensity modulated radiotherapy (t-IMRT) for left-sided breast cancer. It has been shown that MWHD does not necessarily correlate to mean radiation doses, absorbed by cardiac chambers or coronary arteries in adjuvant breast cancer radiotherapy.33,34,35,36 Lately, detailed studies of the specific cardiac structures’ absorbed radiation dose in thoracic radiation therapy24,36,37, and the efforts to understand the specific radiation dose-volume effects in the heart have emerged. 4,38,39,40,41 With expanding knowledge in this field, German Society of Radiation Oncology (DEGRO) recommends new stringent dose constraints for the heart and its substructures: MWHD < 2.5 Gy, left ventricle (LV) D_mean < 3 Gy (LV mean dose), LV V₅ < 17% (volume of LV receiving ≤ 5 Gy), LV V₂₃ < 5% (volume of LV receiving ≤ 23 Gy), left anterior descending coronary (LADCA) D_mean < 10 Gy (LADCA mean dose), LADCA V₃₀ < 2% (volume of LADCA receiving ≤ 30 Gy), and LADCA V₄₀ < 1% (volume of LADCA receiving ≤ 40 Gy).42

To standardize the reporting of cardiac imaging regardless of diagnostic modality, both The American Society of Echocardiography and the European Association of Cardiovascular Imaging recommend using a segmentation model of the LV to assess regional LV function.42,44 The LV segmentation model reflects coronary arteries’ territories and permits to compare echocardiography with other imaging modalities.43 Five main LV segments, defined in a cardiac atlas by Duane et al.24 are based on a previously described 17-segmentation model.44

In this work, we hypothesized that in the setting of the node-negative left-sided breast cancer adjuvant radiotherapy, the lowest median MWHD and doses to the cardiac substructures would be achieved with the t-IMRT, compared to 3D-CRT. In addition, we assumed that individual patient characteristics, which include chest wall separation and breast volume, would contribute to the differences in absorbed doses to the heart and cardiac substructures, regardless of the treatment planning technique. To test our hypothesis, we aimed to quantify doses to the heart and cardiac substructures in present-day free-breathing adjuvant 3D-CRT and t-IMRT and to analyze the differences in dosimetric metrics to organs at risk between three different groups of CTV according to breast size and other individual anatomical information.

By tradition and its contouring pragmatism, MWHD is the most frequently reported surrogate for the assessment of the potential subsequent cardiotoxic effects after radiation therapy for breast cancer. In the present study, we aimed to compare doses to the individual cardiac structures in the circumstances that represent everyday practice in free-breathing node-negative left-sided breast cancer adjuvant 3D-CRT or t-IMRT. Herein, we report reasonably low median MWHD values achieved with both techniques, 1.9 Gy with 3D-CRT and 2.1 Gy with t-IMRT. In the contemporary series, measured mean or median MWHD values in free-breathing node-negative left-sided breast cancer adjuvant RT are in the range of 2.6–3.6 Gy for 3D-CRT6,33,35,36 and 1.8–4.8 for the intensity modulated techniques.11,46,47

In both evaluated techniques, we observed statistically significant differences between the groups of small, medium, and large CTV sizes for the following dose-volume parameters: MWHD, mean doses for proximal LADCA segment, anterior, lateral, inferior, and septal LV walls. In medium and large-sized CTV, we observed reduction of LADCA-D_mean with t-IMRT technique, which was not statistically different. Our results are consistent with previously published studies showing increased CWS, relative to 22 cm, to be one of the predictors for a higher MWHD, in both normo-and hypofractionation.9 Other studies have also demonstrated the correlation between the calculated heart dose and increasing breast size, especially when PTV exceeds 1500 cm³.17,18 Compared to small-sized CTV, MWHD increased with medium-and large- sized CTVs in our study, although the absolute differences between the groups were relatively small, ranging from 0.73 Gy and 0.97 Gy for the t-IMRT and 3D-CRT, respectively. Our results imply that patients’ anatomy, including CWS and/ or CTV/PTV volume, should be also considered when choosing the appropriate radiotherapy technique (3D-CRT vs. modulated approaches), patient setup (prone or lateral vs. supine), and breathing adaptation techniques. As previously mentioned, breast size grouping could be useful in this context, helping to tailor whole breast irradiation.19

Despite the low MWHD for the whole group, our study confirms that apical and anterior parts of LV and mid or distal LADCA segments in both 3D-CRT and t-IMRT techniques receive disproportionately higher D_mean radiation doses. Likewise, in the study by Tang et al., segments corresponding to anterior and apical LV wall absorbed the highest doses, 9.2 Gy and 14.9 Gy, respectively. Patients were treated with tangential breast RT, with or without regional nodal irradiation and with or without DIBH.36 Corresponding values in our study were lower in both evaluated techniques, 3D-CRT vs. t-IMRT for anterior and apical LV walls were 5.0 vs. 5.4 Gy and 8.5 vs. 8.9 Gy, respectively. Lower numbers might reflect a difference in contoured thickness of the LV wall, 6-9 mm in the study of Tang et al. compared to 10 mm used in our study, as suggested by Duane et al.24

In our work, the LADCA-D_mean was 8.2 Gy (range, 1.2–27.9) in 3D-CRT and 8.4 Gy (range, 1.8–27.6) in t-IMRT, respectively. Drost et al. in their systematic review of heart doses reported varying dose-volume measurements for the LADCA. The LADCA-D_mean ranged from 1.9–40.8 Gy (average 12.4 Gy)6, which is similar to our data. In our series of treatment plans, we have demonstrated the highest D_mean for the middle LADCA segment in the group of women with medium-sized CTVs (17.9 Gy) but was not significantly different compared to the smallest or the largest CTV groups. With t-IMRT, it was possible to lower LADCA high-dose areas (V₃₀ Gy), but not low-dose areas or mean doses to the coronary arteries. Carosi et al. observed no difference in MWHD when t-IM-RT was compared to 3D-CRT (2.0 vs. 1.9 Gy) in 24 patients with a median breast volume of 645 cm³. However, the authors showed a statistically meaningful difference in LADCA D_mean (10.3 vs. 11.9 Gy, p=0.0003), LADCA-D_max and LADCA-V₁₇ Gy parameters using t-IMRT compared to 3D-CRT.48

There are many possible explanations for the dissimilar reported heart and heart substructures’ absorbed doses in free-breathing left-breast only RT. The differences may arise from the discrepancy in the total dose prescription and the size of the radiation field, CTV definition and size, OAR contouring, including diameter of the coronary arteries and LV thickness, the lack of detailed heart contouring atlases, individual coronary topology, heart size, body mass index, CWS distance, and finally radiotherapy technique used.9,33,49,50,51,52 The use of contrast agent53 or automatic substructures’ segmentation without54 or with cardiac magnetic resonance imaging55 could improve contouring consistency, but these technical solutions are unlikely to be widely adopted in the near future. Non-automatic contouring is feasible as showed in a study by Francolini et al. Authors made multiple comparisons of delineated cardiac chambers and 5 left LV wall segments according to aforementioned cardiac atlas24 and confirmed high interobserver delineation consistency.56

Spatial variation in contouring has been shown to result in less than 1 Gy dose variation for most segments and in most regimens in adjuvant breast cancer RT, but higher dose variations up to 21.8 Gy were seen for segments close to the radiation field edge.24 Substantial variation in the estimated dose was observed for LADCA, regardless of which particular delineation guidelines were used.57 Except for proximal LADCA (2.6 Gy vs. 2.5 Gy), absorbed mean D_mean values of LADCA segments and LV were lower in our study compared to the partially wide tangential technique used in Duane and coworkers’ research; 15.1 Gy vs. 25.1 Gy for middle LADCA segment, 17.6 Gy vs. 35.8 Gy for distal LADCA segment, and 3.2 vs. 6.7 Gy for LV. In the study of Wennstig et al., three radiation oncologists, using the heart atlas of Feng et al., achieved substantial spatial agreement in delineating coronary arteries on 32 CT study sets. The agreement was the highest for LMCA and LADCA, and less for RCA.23,58 In our study, the coronary vessel diameter was set to 6 mm considering both cardiac and respiratory motion, similar to Wennstig and colleagues’ work.23

Based on recent clinical reports, the DEGRO group proposed stringent dose constraints for the heart and its substructures in adjuvant breast cancer radiation treatment.42 We surpassed at least one of the proposed optimal dose constraints for LV (D_mean < 3 Gy, V₅ < 17%, and V₂₃ < 5%) or LADCA (D_mean < 10 Gy, V₃₀ < 2%, and V₄₀ < 1%) in 11.7–51.7% of all evaluated plans. In our plan optimization process, we did not use specific dose-volume constraints for cardiac substructures. However, it has been shown that additional LADCA or LV constraints in breast cancer adjuvant 3D-CRT or IMRT treatment planning might help to optimize heart dosimetric metrics further.23,59

In our study, the evaluation of the planned dose to the heart and specific cardiac substructures was performed in a free-breathing simulation CT scan and in the supine position. Ideally, the dose to cardiac substructures should also be evaluated for patients treated using alternative treatment positions (lateral decubitus or prone) or with DIBH. Due to various reasons, most patients are still treated in the conventional free-breathing supine position, whereas prone positioning or DIBH is in the best-case scenario offered to only 28–83% of breast cancer patients.15,60 All delineations were performed on a non-enhanced CT scan, an approach that may impact the visibility of the small cardiac segments. Additional drawback of our study is not including patients receiving peri-clavicular regional nodal irradiation with or without internal mammary lymph chain irradiation. Strengths of this study include careful contouring of individual cardiac substructures and using a cardiac atlas based on individual anatomy. An experienced cardiac radiologist thoroughly evaluated the contours.

This is the first study to evaluate the cardiac contouring atlas for radiotherapy by Duane et al.24 simultaneously considering different CTV size. We confirmed that regardless of very low D_mean values for the whole heart achieved using a 3D-CRT or t-IMRT free-breathing adjuvant RT technique for breast cancer, a small volume of the heart may receive disproportionate D_mean or D_max values exceeding 40 Gy. We observed differences in heart dosimetric metrics between the small, medium, and large CTV sizes for both evaluated techniques, which may disappear with DIBH technique. With t-IMRT technique, only few dosimetric metrics were improved compared to 3D-CRT. The observed results in our study suggest that anatomic differences, especially breast volume and CWS, should be considered in clinical practice as well as in the dosimetric studies of various treatment planning techniques. Subdividing breast target volume into similar cohorts could be helpful in this context and further research is warranted. The quantification of the radiation dose variability of individual cardiac substructures is an important first step to understand the unique cardiac structures’ dose-volume predictors for cardiotoxicity in adjuvant, free-breathing breast cancer radiation therapy. In the future, reported absorbed doses may be paired with cardiac imaging and help to choose patients for whom more intense cardiac function monitoring is warranted.