In this study, we identified two structural metrics on CT scans of left-sided breast cancer patients with modified radical mastectomy and investigated the relationships between these metrics and the heart and left lung absorbed dose, including the mean dose of heart and left lung, V5, V30 of heart and V5, V10, V20, V30, V40 of left lung. To the best of our knowledge, this study was the first research, evaluated these structural metrics and relationship between them and the heart and left lung dose, to present.
According to the results, CI index was associated strongly with all of the dose metrics of heart in 3D-CRT plans, the correlation coefficients were 0.857 (p༜0.01), 0.814 (p༜0.01), and 0.869 (p༜0.01) of MHD, V5, and V30 of heart, respectively. Recently, Cao et al. indicated that the cardiac contact distance (CCDps) showed a positive linear correlation with the MHD (r = 0.63, p༜0.01) in their study[16]. They also suggested that the lateral heart-to-chest distance (HCD) demonstrated a negative linear correlation with the MHD (r=-0.65, p༜0.01). Similarly, Mendez et al. investigated predictors (4th Arch and 5th Arch) in another study, involved a simple linear line drawing to the 4th or 5th costal arch level (4th Arch, 5th Arch), from the left edge of the sternum to the anterior portion of the left lung[17]. The previous study shown that the correlation coefficient of 4th Arch and MHD was 0.61 (p༜0.05), while 4th Arch and V25 of heart was 0.57 (p༜0.05). Despite its reasonable prediction capacity, it was not clear if the CT scan would accurately acquire slice at the level of the 4th costal arch because the thickness of the 4th costal arch is well beyond the range of the thickness of CT scans. In our research, we acquired the structural metrics through the entire range of the organs at risk, so we could ignore the CT slice thickness. Other studies had also shown that the maximum heart distance to the chest wall correlates with mean dose of heart[18–20]and can reliably estimate cardiac exposure in patients treated with breast RT. In line with the previous studies, our results implied that the dose distribution of heart greatly dependent on the proximity of the heart to the irradiation fields. Nevertheless, we only observed moderate correlation between CI index and the dosimetry of heart in VMAT plans. We hypothesized that the dose delivered by the arc of radiation fields would render less high dose to the heart in a nonlinear fashion, thus the correlation was weaker in VMAT compared to 3D-CRT.
Interestingly, the intercept of the two formulas for MHD in 3D-CRT and VMAT plans was when CI index equaled 0.042, the MHD equaled 512.33 cGy, which was slightly less than the mean dose of heart in both plans. Considering the slop of formulas of 3D-CRT plans was steeper than formulas of VMAT plans, the MHD of 3D-CRT plans would surpass the one of VMAT plans when CI index was more than 4.2% (Fig. 2.). Therefore, we recommend that VMAT plan is preferable when CI index is greater than 4.2%, and 3D-CRT plan is the first consideration when CI index is less than 4.2%.
In this study, we examined the influence of PI index on dose metrics of left lung as well. There were strongly positive associations between PI index and absorbed dose parameters in 3D-CRT and VMAT plans as illustrated in Fig. 3. Although the linear regression formulas for MLLD in both 3D-CRT and VMAT plans intercepted at PI index equaled 0.063 and MLLD equaled 1180.46 cGy, we supposed this is not significantly clinically meaningful, for the reason that breast cancer with supraclavicular region would produce PI index more than 6.3% for most of the time. Since the slops of formulas in 3D-CRT plans are steeper than those of VMAT plans, the greater the PI index the greater the absorbed does to the left lung in 3D-CRT plans compared to VMAT plans as a result. In addition, the MLLD of 3D-CRT plans was 1420.31 cGy, and its corresponding PI index was 14.6%. Thus, we suggest that when PI index is greater than 14.6%, we could choose VMAT plan firstly, yet we may choose 3D-CRT plan in the opposite.
We also found that VMAT plans exhibited advantageous dosimetry in high-dose compared with the 3D-CRT plans in the current study. The V30 of heart and V30, V40 of the ipsilateral lung were considerably lower in VMAT plans than in 3D-CRT plans (p༜0.01). Moreover, the mean dose of ipsilateral lung in VMAT plans were less than in 3D-CRT plans as well, in accordance with the studies of Liu et al. and Mo et al[14, 15], but opposite to the results of Bogue’s study[13]. Incidentally, we discovered that the PI were larger in the VMAT plans than in 3D-CRT plans. Although we observed a similar mean dose of heart in either cohort plans in our research, the larger PI in VMAT may indicated that VMAT plans played an important role in sparing the heart from redundant doses. However, the V5 of heart, V5, V10 of ipsilateral lung, and V5, V20 and mean dose of contralateral lung, as well as the V5 and mean dose of contralateral breast were higher in the VMAT plans due to the low dose spread of VMAT in our study, the results agreed with many studies comparing the dose of VMAT and 3D-CRT plans[13–15].
There are several limitations in this planning study. First, we recognize that the prediction of heart and lung dose is sophisticated and using only structural metrics may be incomplete. Second, there is no standard VMAT, and the dose in OAR depend on widely varying technology, beam setup, OAR constraints et al. Thus, the relation between structural metrics and dose distributions may lack of generalizability. Third, although CI index is similar between groups, it should be noted that the PI index is different between these cohorts, suggesting that the groups are not completely matched, the VMAT group may have larger or longer target.