In this study, parameters were explored that could potentially guide the clinical decision regarding whether application of the DIBH technique is likely to reduce the dose to the heart in patients with left-sided breast cancer, thus potentially reducing the risk of development of chronic heart disease in the future. Moreover, we developed a model that could help to identify these patients before initiation of radiotherapy treatment planning.

In clinical routine, any benefit from DIBH irradiation is evaluated during the radiotherapy plan comparison between the plans in free breathing and in DIBH [20, 22], which is time consuming with respect to the whole planning process. The model developed here does not only consider physical parameters in terms of constraints of organs at risk but also clinical parameters of the individual patients. Parameters with a direct influence on DIBH irradiation are tumor localization, the volume of the left lung, and the medial distance from the chest wall to the heart, while age and ECOG performance status are factors that may influence the capability of the patient to perform DIBH. In addition, the DIBH technique also requires patients to constantly hold their breath and avoid abdominal breathing.

This analysis showed that fewer patients received DIBH with increasing age. It must be considered that some treatment-related heart toxicities only develop after 10 to 15 years [1]. In patients with a lower life expectancy, irradiation in free breathing may therefore lead to reduction of local recurrences and tumor-dependent overall survival without increasing heart toxicity.

Patients with tumor localization in inner quadrants were more often assigned to the DIBH subgroup, which is likely due to the close distance between the chest wall and the heart and, thus, to a higher heart dose. This is in line with Bouchardy et al., who found a higher cardiac mortality in patients with lesions within the inner quadrants of the left breast [23]. Moreover, patients with smaller lung volumes are more likely to be assigned to the DIBH subgroup. Here, less lung volume between the chest wall and the heart leads to higher doses to the heart. This is possibly why the volume of the left lung was considered instead of the total lung volume in the model. However, both parameters were strongly correlated.

It is well known that the individual patient anatomy influences the heart dose and consequently cardiac toxicity [24, 25]. In several studies, potential anatomical parameters were analyzed for predicting the mean heart dose before initiation of treatment planning, e.g., in [22, 26, 27]. Lee et al. showed an association with the number of CT slices in which the heart was close to the chest wall [26], whereas Hiatt et al. measured the contact length between the heart and the anterior chest wall [27]. In Cao et al., a model was developed to predict the reduction of heart and lung doses in DIBH irradiation [22], with the lateral heart-to-chest distance/cardiac contact distances in parasagittal CT planes as strong predictors. Here, the medial distance between the outer chest wall was selected in the model, although there was a strong correlation between both the medial and the lateral distance (Spearman correlation 0.57), suggesting that both are indicative of benefits from DIBH irradiation. Previous or parallel irradiation of the contralateral breast was not selected by the model. Moreover, the application of systemic therapy seemed to be less important for assigning patients to the subgroups.

Overall, DIBH irradiation led to significantly lower mean (mean ± standard deviation DIBH: 2.0 ± 1.0 Gy vs. free breathing: 3.4 ± 1.6 Gy; p < 0.001) and maximum (30.8 ± 13.9 Gy vs. 40.6 ± 10.7 Gy; p < 0.001) doses to the heart and maximum doses to the LAD (17.1 ± 13.1 Gy vs. 25.5 ± 16.0 Gy; p < 0.001). Coronary disease, including reduced perfusion, mainly occurs more than 10 years after irradiation [8, 25, 28]. Therefore, the LAD might be a substructure for dose reduction. However, due to its small volume, decreased visibility, and resultant contouring uncertainties in CT scans [29], it remains unclear whether it is superior to the mean dose of the heart or any other substructure.

Taken together, the model can identify patients who receive lower heart doses through application of DIBH; however, it warrants validation in an independent dataset. The extent of the clinical benefit for patients cannot be concluded from our study, as this would require a much larger patient cohort with long-term clinical outcomes. Further limitations are the retrospective monocentric nature of the study, missing randomization, and exclusion of patients for DIBH planning CT scans with the mean heart dose up to 5 Gy criterion already met (according to the internal SOP relevant at the time). The latter also adds to the bias of the preselection of patients for DIBH planning CT, together with other clinical parameters potentially affecting the clinician’s decision, such as application of chemotherapy. In addition, one cannot rule out that patients with a mean heart dose of less than 5 Gy might also develop chronic heart disease after irradiation and, thus, also benefit from DIBH. Also, patients with lymph node irradiation other than in axilla levels I and II were not considered. In conclusion, the model is easy to apply and may potentially guide the clinician’s decision for the individual patient receiving radiotherapy of the left breast regarding the prescription of a DIBH planning CT, thus also saving time in the planning process. Future investigations including outcome data are needed to prove the patient benefit in terms of reduction of chronic heart toxicity through application of DIBH irradiation.